UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Manganese imaging with positron emission tomography, autoradiography, and magnetic resonance Topping, Geoffrey John 2014

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2014_spring_topping_geoffrey.pdf [ 10.39MB ]
Metadata
JSON: 24-1.0165883.json
JSON-LD: 24-1.0165883-ld.json
RDF/XML (Pretty): 24-1.0165883-rdf.xml
RDF/JSON: 24-1.0165883-rdf.json
Turtle: 24-1.0165883-turtle.txt
N-Triples: 24-1.0165883-rdf-ntriples.txt
Original Record: 24-1.0165883-source.json
Full Text
24-1.0165883-fulltext.txt
Citation
24-1.0165883.ris

Full Text

Manganese Imaging with Positron Emission Tomography,Autoradiography, and Magnetic ResonancebyGeoffrey John ToppingA THESIS SUBMITTED IN PARTIAL FULFILMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinThe Faculty of Graduate and Postdoctoral Studies(Physics)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)February 2014? Geoffrey John Topping, 2014iiAbstractManganese is a magnetic resonance imaging (MRI) contrast agent for small animals that can provide a blood-flow-independent measure of neuronal activation. Established Mn MRI methods have limited ability to measure concentrations of Mn or details of its distribution in vivo, which limits theirexperimental power. Positron emission tomography (PET) can measure quantitative distributions in vivo in small animals of positron-emitting radionuclides such as Mn-52, although has poorer spatial resolution than MRI. Autoradiography (AR) can also measure quantitative distributions of Mn-52, in ex vivo brain tissue, with spatial resolution similar to MRI.This work has three primary goals: to develop and characterize Mn-52 as a radionuclide for PET in phantoms and in small animals, to develop a quantitative MRI method for measuring Mn concentration in the brain of small animals, and to validate the MRI results by comparisons with AR and PET.Mn-52 was produced by proton irradiation of natural Cr foil, isolated by column chromatography, and used as a PET tracer for the first time in phantoms and in vivo in rats. Mn-52 phantom image quality metrics were similar to F-18, an established PET radionuclide. After systemic administration in rats, Mn-52 accumulation was seen in bones, but little was seen in the brain, due to the blood-brain barrier. Direct injection into the lateral ventricle effectively delivers Mn-52 throughout the rat brain. Mn-52 AR images were acquired and used for comparison with MRI.MRI R1 relaxation rate maps of rat brain were acquired using a radiofrequency field strength independent inversion recovery Look-Locker sequence, and used to generate relaxation rate change and Mn concentration maps after Mn administration. These maps showed excellent quantitative agreement with PET and AR images of the same animal, confirming that MR R1 change accurately measures Mn concentration in rat brain in the range 0 to 0.1 mM, in the absence of other sources of R1 change. However, at some point above this concentration, measured R1 becomes inaccurate. Accordingly, Mn concentration mapping with MRI is a potentially useful tool to improve the experimental power of Mn-uptake imaging to assess neuronal activation.iiiPrefaceThe work presented in this thesis was conducted in the University of British Columbia (UBC) Hospitalat the UBC Point Grey campus, in the UBC MRI Research 7T Facility at the UBC Point Grey campus, and in the Nuclear Medicine Laboratories at TRIUMF.Radioisotope production and radiochemistry were reviewed by the TRIUMF Safety Committee (project number LS-94) and the TRIUMF Life Sciences Projects Evaluation Committee (LSPEC).Animal experiments were conducted under protocols approved by the UBC Animal Care Committee (application numbers A10-0142, A10-0133, and A09-0438).Part of tracer production and PET imaging discussed in chapters 2 and 3 were published (G.J. Topping, P. Schaffer, C. Hoehr, T.J. Ruth, V. Sossi, "Manganese-52 positron emission tomography tracer characterization and initial results in phantoms and in vivo", Medical Physics 40, 042502-1-042502-8 (2013)).Geoffrey Topping was the lead investigator, and was responsible throughout the project for experiment design, acquiring ethics and safety approvals, experiment scheduling and facility booking, performing radiochemistry, purchasing supplies and materials, arranging for commercial chemical analysis, operating the PET scanner, animal monitoring, writing analysis and reconstruction software, PET data reconstruction, data analysis, and manuscript composition, submission, and revision. He also contributed to operation of the MRI scanner for data acquisition in chapter 6, and animal monitoring and maintenance during MRI scanning in chapter 7.Vesna Sossi was the primary research supervisor, contributed to experiment design, team formation,and project supervision for all experimental work, and for chapters 2, 3, and 4 in particular.Paul Schaffer contributed to radionuclide production design and radiochemistry process design and related safety approvals in chapter 2.ivCornelia Hoehr contributed to design for radionuclide production by irradiation in chapter 2.Thomas Ruth contributed to radiochemistry process design in chapter 2.Linda Graham and Wade English prepared Cr foil for irradiation, performed irradiation for Mn-52 production, and extracted irradiated foil for tracer production in chapter 2, and performed irradiation for F-18 PET scans in chapter 3.Mike Adam performed radiochemistry for F-18 PET scans in chapter 3.Ivan Klyuzhin constructed part of a phantom object used in chapter 3.Greg Stortz assisted in collected of some PET phantom images used in chapter 3.Siobhan McCormick routinely acquires PET normalization data which was used in chapter 3.Rick Kornelsen handled animals during PET experiments and contributed to animal monitoring in chapters 3 and 4, and prepared brain slices, exposed autoradiographic plates, and operated the autoradiographic read out equipment for AR experiments in chapter 4.Katie Dinelle assisted with scheduling of irradiations and experiments in chapters 2, 3, and 7.Chenoa Mah contributed to animal handling during PET and before MRI experiments in chapters 3 and 7.Andrew Yung contributed to MRI experiment planning and design, writing MRI acquisition and image analysis software, MRI coil and other apparatus design and construction, animal handling, experimental setup, and operating the MRI scanner in chapters 6 and 7.Piotr Kozlowski contributed to MRI experiment design in chapters 6 and 7.vBarry Bohnet contributed to animal handling, experimental setup, and operating the MRI scanner in chapter 7.viTable of ContentsAbstract ........................................................................................................................................... iiPreface ............................................................................................................................................ iiiTable of Contents ............................................................................................................................ viList of Tables .................................................................................................................................... xiiList of Figures .................................................................................................................................. xiiiGlossary ........................................................................................................................................... xxxAcknowledgements ......................................................................................................................... xxxiiiMotivation ..................................................................................................................................... .. xxxiv1 Positron Emission Tomography Introduction ............................................................................... 11.1 Source of PET Images ................................................................................................... 11.1.1 Radioactive Decay ........................................................................................ 11.1.2 Positron Decay ............................................................................................. 21.1.3 Positron Annihilation .................................................................................... 41.2 PET System ................................................................................................................... 51.2.1 Scintillators ................................................................................................... 61.2.2 Photomultipliers ........................................................................................... 71.2.3 Event Characterization ................................................................................. 81.3 Reconstruction ............................................................................................................. 111.3.1 Forward Projection ....................................................................................... 111.3.2 Sinograms ..................................................................................................... 121.3.3 Filtered Back Projection ................................................................................ 141.3.4 Other Reconstruction ................................................................................... 161.4 Corrections ................................................................................................................... 171.4.1 Randoms ....................................................................................................... 171.4.2 Attenuation .................................................................................................. 181.4.3 Scatter .......................................................................................................... 211.4.4 Cascade Coincidences .................................................................................. 231.4.5 Normalization ............................................................................................... 25vii1.4.6 Calibration .................................................................................................... 281.5 PET Experiments ........................................................................................................... 291.5.1 Biology and Tracers ....................................................................................... 291.5.2 PET Image Analysis ....................................................................................... 302 Tracer Production ......................................................................................................................... 332.1 Background .................................................................................................................. 332.1.1 Irradiation ..................................................................................................... 332.1.2 Chemistry ..................................................................................................... 352.2 Mn-52 ........................................................................................................................... 362.2.1 General ......................................................................................................... 362.2.2 Cascade Gammas ......................................................................................... 372.3 Mn-52 Production ........................................................................................................ 392.3.1 Irradiation ..................................................................................................... 392.3.2 Yield Measurements ..................................................................................... 412.3.3 Mn-54 Contamination / Waste Storage ........................................................ 432.3.4 Specific Activity ............................................................................................ 452.4 Mn-52 Energy Spectra .................................................................................................. 462.4.1 Acquisition .................................................................................................... 462.4.2 Results .......................................................................................................... 472.5 Mn-52 Radiochemistry ................................................................................................. 482.5.1 Column Chromatography ............................................................................. 492.5.2 Resin Selection ............................................................................................. 502.5.3 Cr Content Testing ........................................................................................ 522.5.4 Separation Procedure ................................................................................... 532.5.5 Separation Results ........................................................................................ 553 Positron Emission Tomography Imaging ....................................................................................... 563.1 Focus 120 microPET Scanner ........................................................................................ 563.2 Phantom Imaging ......................................................................................................... 573.2.1 Initial Phantom ............................................................................................. 58viii3.2.2 Cascade Correction ....................................................................................... 623.2.3 Image Quality Phantom ................................................................................ 703.2.4 Resolution Phantom ..................................................................................... 733.2.5 Calibration Phantom ..................................................................................... 743.2.6 Contrast Phantom ........................................................................................ 773.2.7 Normalization Phantom ............................................................................... 783.2.8 Further Normalization Testing ...................................................................... 803.3 Animal Mn-52 Imaging ................................................................................................. 823.3.1 General ......................................................................................................... 823.3.2 First Mn-52 IP Injection ................................................................................ 853.3.3 First Mn-52 IV Injection ................................................................................ 863.3.4 Second Mn-52 IV Injection ........................................................................... 883.3.5 IV F-18 Water Injection ................................................................................. 943.3.6 ICV Mn-52 Injections .................................................................................... 963.4 Animal Mixed Mn-52 and Nonradioactive MnCl2 Imaging ............................................ 993.4.1 Mixed ICV Injection ...................................................................................... 993.4.2 Second Mn-52 IP Injection ........................................................................... 1013.5 Discussion ..................................................................................................................... 1054 Autoradiography ........................................................................................................................... 1074.1 Background .................................................................................................................. 1074.2 Calibration .................................................................................................................... 1094.3 Preparation and Acquisition ......................................................................................... 1104.4 Background Correction ................................................................................................. 1144.4.1 Raw Data ...................................................................................................... 1154.4.2 Constant Subtraction .................................................................................... 1174.4.3 Position-Dependent Subtraction .................................................................. 1194.4.3.1 IP Injection .................................................................................... 1194.4.3.2 ICV Injections ................................................................................ 1214.5 Slice Registration .......................................................................................................... 1254.6 Results .......................................................................................................................... 129ix5 Magnetic Resonance Imaging Background .................................................................................. 1325.1 Physical Basis ................................................................................................................ 1325.1.1 Nuclear Spins and Macroscopic Magnetization ........................................... 1325.1.2 Signal Reception ........................................................................................... 1355.1.3 Precession .................................................................................................... 1365.1.4 Relaxation ..................................................................................................... 1385.2 Radio-frequency Pulses ................................................................................................ 1425.2.1 Physical Description ...................................................................................... 1435.2.2 Signal Formation ........................................................................................... 1455.2.2.1 90 Degree Pulse ............................................................................ 1455.2.2.2 T2* Contrast ................................................................................. 1455.2.2.3 T1 Contrast ................................................................................... 1465.2.2.4 180 Degree Inversion .................................................................... 1475.2.2.5 T2 Weighting / Spin Echo .............................................................. 1485.2.2.6 Inversion Recovery ....................................................................... 1515.2.2.7 Fast Repetition Time ..................................................................... 1525.3 Spatial Localization ....................................................................................................... 1555.3.1 Frequency / Phase Encoding ........................................................................ 1555.3.2 Slice Selection ............................................................................................... 1575.4 T1 Mapping .................................................................................................................. 1585.4.1 Inversion Recovery ....................................................................................... 1585.4.2 Variable Flip-Angle Steady State ................................................................... 1595.4.3 Look-Locker .................................................................................................. 1625.4.3.1 Look-Locker Signal Dependence ................................................... 1625.4.3.2 Flip-Angle Independence .............................................................. 1656 Manganese-Enhanced Magnetic Resonance Imaging .................................................................. 1686.1 Manganese Uptake ....................................................................................................... 1686.2 MEMRI Applications ..................................................................................................... 1696.3 Paramagnetic Relaxation .............................................................................................. 170x6.4 MEMRI Complications .................................................................................................. 1726.4.1 Blood Brain Barrier ....................................................................................... 1726.4.2 Manganese Toxicity ...................................................................................... 1736.5 Mn Concentration Mapping ......................................................................................... 1746.5.1 Relaxivity Calibration .................................................................................... 1746.5.2 Other Calibration Factor Limitations ............................................................ 1796.5.3 Discussion ..................................................................................................... 1807 Magnetic Resonance Imaging ...................................................................................................... 1817.1 7T MRI System .............................................................................................................. 1817.2 R1 Relaxation Rate Mapping ........................................................................................ 1837.3 Initial IP MnCl2 Injection ............................................................................................... 1857.4 ICV MnCl2 Injections ..................................................................................................... 1887.4.1 ICV Preparations ........................................................................................... 1897.4.2 ICV Imaging ................................................................................................... 1917.5 Mixed Mn-52 and Nonradioactive Mn Injections ......................................................... 1947.5.1 Mixed Mn ICV Injections .............................................................................. 1957.5.2 Mixed Mn IP Injections ................................................................................. 2017.6 Relaxation Rate Map Analysis ...................................................................................... 2047.6.1 Coregistration ............................................................................................... 2057.6.2 R1 Difference Maps ...................................................................................... 2067.7 Discussion ..................................................................................................................... 2088 Multimodality Comparisons ......................................................................................................... 2098.1 Multimodality Correlation ............................................................................................ 2098.1.1 Registration .................................................................................................. 2108.1.2 Comparison .................................................................................................. 2128.2 Multimodality Calibration ............................................................................................ 2178.3 Mn Dose Effects ............................................................................................................ 2228.4 Discussion ..................................................................................................................... 227xi9 Discussion and Conclusions .......................................................................................................... 2289.1 Future Work ................................................................................................................. 2289.1.1 Mn-52 PET With Activation-Induced Uptake ................................................ 2289.1.2 Blood Brain Barrier Disruption ..................................................................... 2289.1.3 Bone Imaging ................................................................................................ 2299.1.4 Improved Tracer Production ......................................................................... 2299.1.5 Additional Calibration Comparisons ............................................................. 2309.1.6 Mn MRI Limitations ...................................................................................... 2309.1.7 Mn MRI Concentration Mapping With Activation-Induced Uptake ............. 2319.2 Conclusions .................................................................................................................. 232References ....................................................................................................................................... 235Appendix A Imaging Subjects .......................................................................................................... 245xiiList of Tables1.1 Theoretical ranges of travel in water before positron annihilation after emission from PETradionuclides (Le Loirec, 2007). ............................................................................................ 51.2 511 keV photon linear attenuation factors (?) for selected materials from the microPET Focus reconstruction software. These factors account for all interaction mechanisms that contribute to photon attenuation. ........................................................................................ 201.3 High probability Mn-52 decay cascade gamma emissions (NNDC Database). ...................... 243.1 Mn-52 and F-18 activity recovery coefficients (RC) in small diameter cylinders, relative to uniform region of image quality phantom. ........................................................................... 727.1 Intra-cerebro-ventricular (ICV) injections of MnCl2 in water, phosphate-buffered saline (PBS), or PBS mixed with chromatography column eluent residue, into rats, and observations of reactions. .................................................................................................... 1908.1 Linear regression fit statistics for R1 change plotted against autoradiography counts. ....... 2168.2 Concentration and specific uptake values in PET images in ROIs on brain and pituitary glands of rats after systemic injection of Mn-52 with or without additional MnCl2 mixed into the injection solution. ....................................................................................................225xiiiList of Figures1.1 General Electric (Connecticut, USA) Advance human PET system with ring of detector blocks visible. ........................................................................................................................ 61.2 Example arrangement of 7x7 grid of crystals (black) coupled to 4 photomultipliers (red). Actual PET systems will in general use different packing geometries than shown here for the photomultipliers and relative sizes of the crystals and photomultipliers. ...................... 71.3 Annihilation site (green), and paths of anti-parallel annihilation photons (blue), to the surrounding ring of PET detectors (red). The PET system sees only which detector elements in which the photons are detected but knows that the annihilation likely occurred in the field of view along the line connecting those elements. ............................. 91.4 Example single plane cut across the field of view of a PET scanner (enclosed by red circles) containing a source distribution (solid black), and a single angular projection (at right) of that source distribution, as measured by the PET system. ..................................... 121.5 Single slices of sinogram data. Slices images both have horizontal axes in the radial direction. The left image has the vertical axis in the circumferential direction, and shows the characteristic sine-wave shape from which sinogram get their name. The right image has vertical axis in the scanner axial direction, and shows an axial series of projections from the same angle. ............................................................................................................ 131.6 Slice of PET images of 1.1 mm inner diameter capillary tube containing PET tracer. Image scale was adjusted to emphasize the appearance of radial streak artifacts in FBP reconstruction. ...................................................................................................................... 151.7 Profiles through image of uniform phantom of 7 cm diameter containing F-18 reconstructed with and without attenuation correction, with profile values scaled to be similar at edges where attenuation effects are minimal. The uncorrected image has distinctly lower values near the middle, where activity is surrounded by the most attenuating material. ............................................................................................................ 211.8 Normalization factors sinogram. Measured data sinograms are rescaled bin-by-bin duringreconstruction to compensate for these normalization factors. .......................................... 271.9 Raw sinogram data before normalization factors are applied. ............................................. 27xiv1.10 Sinogram data after normalization factors are applied, with substantially reduced patternof streaks. ............................................................................................................................. 282.1 Decay scheme of Mn-52 (NNDC MIRD). ............................................................................... 382.2 Irradiated Cr foil piece in beaker. Courtesy Paul Schaffer (TRIUMF). .................................... 402.3 Gamma spectrometer shielded sample container and coolant supply. ................................ 422.4 Energy spectra of Mn-52 as measured with multichannel gamma spectrometer and with the microPET Focus 120 system's adjustable energy window, and spectrum of F-18 measured with microPET. ..................................................................................................... 472.5 Beaker of Cr foil dissolved in HCl on hot plate. ..................................................................... 482.6 Two chromatography columns eluting green Cr, with resin bed above with bulk of Cr passed, and Cr at top filtering through the column to drive elution. ................................... 492.7 Mn-52 activity (curves) dependence on elution fraction number for Dowex and BioRad resins from similarly prepared columns. Also shown are qualitative assessments of the colour of the elutions, which roughly indicate Cr content of the solution, with darker green colour indicating more Cr, and paler yellow colour indicating less Cr content. The BioRad resin has better separation between the peak of the Mn-52 activity, and the range of fractions with visible indication of Cr content. ....................................................... 512.8 Plots of Cr content and Mn-52 activity in similarly-prepared elution fractions. Some overlap of Cr and Mn content is seen, but Cr is largely absent by the peak of the Mn content curve. ....................................................................................................................... 533.1 The microPET Focus 120 scanner, with 30 ml phantom bottle being scanned. .................... 573.2 500 ml Nalgene bottle containing non-radioactive tap water, and 30 ml bottle containing Mn-52. Bottle is on the microPET scanner bed in preparation for a scan. ........................... 583.3 Single transaxial slice of 500 ml Nalgene water bottle and 30 ml bottle of Mn-52 solution.Data were acquired at 350-750 keV energy window, and reconstructed without cascade correction. The central orange circle is the 30 ml bottle, and the bluish solid shape around it is the outline of non-radioactive water that has been rendered visible by the reconstruction not accounting for cascade background. Colour bar scale ranges from 0 to 100% relative image intensity. .............................................................................................. 59xv3.4 Single transaxial slice of 500 ml Nalgene water bottle and 30 ml bottle of F-18 solution. Data were acquired at 350-750 keV energy window. The central orange circle is the 30 mlbottle, and no discernible background appears around the bottle, unlike with Mn-52 at the same energy window. Colour bar scale ranges from 0 to 100% relative image intensity. ............................................................................................................................... 603.5 Single transaxial slice of 500 ml Nalgene bottle and 30 ml bottle of Mn-52 solution. Data were acquired at 450-600 keV and reconstructed without cascade correction. The restricted energy window has reduced the appearance of the background compared withthe previous figure. Colour bar scale ranges from 0 to 100% relative image intensity. ............................................................................................................................... 623.6 Angle-averaged sinogram profiles for a single segment and axial offset. The uncorrected sinogram profile, the rescaled randoms profile, and the cascade-corrected angle-averaged profile are shown. ................................................................................................. 633.7 Single scanner-axial-position slices from sinograms of PET data acquired on the Focus 120 microPET system. Plot axes are the histogram bin radial offset (horizontal) and circumferential projection angle (vertical).  Scale units arbitrary. Top: The 30 ml phantom bottle filled with F-18 with a 450-600 keV energy window. Centre: A rat's head after IP injection of F-18, scanned with a 350-750 keV energy window. Bottom: A rat's abdomen after IP injection of F-18, scanned with a 350-750 keV energy window. .............................. 643.8 Profiles through sinograms of 30 ml bottles containing F-18 (top, centre) and Mn-52 (bottom). Data were acquired with energy windows of 350-750 keV (top), and 450-600 keV (centre, bottom). Profiles were generated by averaging five sinogram planes (each similar to the top figure in the previous set) and then extracting a single row (representing a single angular projection). For each profile, 20 central peak values were averaged, and used to normalize the profile to have a peak near 1, so that relative scale of backgrounds could be compared. .................................................................................... 653.9 Scale factors matching peripheral sinogram bins of estimated randoms distribution to measured coincidence data, plotted against sinogram plane number (axis labelled "Sinogram Bin") for first 95 sinogram planes, which are the planes that are perpendicularto the scanner axis. Shown are data for a phantom (left) and rat brain (right) containing Mn-52. .................................................................................................................................. 68xvi3.10 Single transaxial slice of 500 ml Nalgene bottle and 30 ml bottle of Mn-52 solution. Data were acquired at 450-600 keV and reconstructed with cascade correction. Appearance ofbackground outside the central peak region is reduced by the correction. Colour bar scale ranges from 0 to 100% relative image intensity. .......................................................... 683.11 Radial profiles through 500 ml Nalgene bottle and 30 ml bottle containing Mn-52 solution. Profiles are shown at energy windows of 350-750 keV without cascade correction, and at 450-600 keV without and with cascade correction applied before reconstruction. The combination of the restricted energy window and cascade correctionis very effective at removing cascade background in reconstructed images. .................... 693.12 Axial slices from PET images of image quality phantom filled with Mn-52 (top) and F-18 (bottom). Uniform region (left), 1 to 5 mm diameter cylinders (centre), and water-filled (left side of right images) and air- filled (right side of right images) enclosures are shown. Both sets of slices are shown scaled relative to their own peak values. Colour bar scale ranges from 0 to 100% relative image intensity. ................................................................... 703.13 Images of Mn-52 (left) and F-18) capillary tubes as resolution phantoms in microPET. Profiles are shown placed through the central peaks. Colour bar scale ranges from 0 to 100% relative image intensity. .............................................................................................. 743.14 Single transaxial slice of Mn-52 calibration phantom with ROI drawn. Colour bar scale ranges from 0 to 100% relative image intensity. ................................................................... 753.15 Natural logarithm of reconstructed activity concentration on Mn-52 phantom scanned repeatedly. ............................................................................................................................ 763.16 Single transaxial slice of PET image of two glass vials containing different concentrations of Mn-52. ROIs are drawn on each vial. Colour bar scale ranges from 0 to 100% relative image intensity. ..................................................................................................................... 773.17 The same single coronal slice through PET images reconstructed from the same acquired data of calibration phantom filled with Mn-52, reconstructed with normalization data generated with F-18 (left) and Mn-52 (right). What appears to be a normalization artifact appears under the black + in the Mn-52-normalized image. Colour bar scale ranges from 0 to 100% relative image intensities. .................................................................................... 79xvii3.18 Profiles along the axial direction of the same single coronal slice through PET images reconstructed from the same acquired data of calibration phantom filled with Mn-52, reconstructed with normalization data generated with F-18 and Mn-52. ............................ 793.19 Axial (top) and radial (bottom) profiles on near-central planes of PET images of bottles of F-18 and Mn-52. ................................................................................................................... 813.20 Rat on microPET scanner bed. Pulse oximeter and rectal thermometer are in place. Snoutis within a makeshift anesthetic nose cone used for this scan instead of the plastic cone used when the head-holder is mounted. .............................................................................. 833.21 First rat IP injection Mn-52 microPET images (red-white), overlaid on attenuation maps (grayscale background) to provide some anatomical context. Transaxial (left), coronal (middle) and sagittal (right) slices are shown. Images were acquired 21 (top) and 45 (bottom) hours after IP injection. The PET images are not cascade corrected, and were not reconstructed with calibrated normalization data, and were smoothed to improve visibility of features seen here. ............................................................................................. 863.22 Sagittal (top) and coronal (bottom) slices through PET image of rat body after Mn-52 IV injection. Image is a composite of 4 acquisitions of approximately 90 min each, after 5 min biodistribution post-injection. This image was cascade corrected, but not calibrated. ............................................................................................................................. 883.23 Time activity curves (TACs) of abdominal regions of interest (ROIs) in Mn-52 image of rat from 0 to 30 min post-injection of Mn-52. ........................................................................... 893.24 Maximum intensity projections through abdomen of rat of PET images acquired after IV injection of Mn-52. Images are shown from 0-30 min (left) and 60-75 min (right). ............. 903.25 Coronal (left) and sagittal (right) slices of PET image of rat acquired for 30 min immediately after IV injection of Mn-52. Isotropic Gaussian smoothing with FWHM of 1 mm was applied. ................................................................................................................... 913.26 Coronal (left) and sagittal (right) slices of PET image of rat 6 days after IV injection of Mn-52. Isotropic Gaussian smoothing with a FWHM of 1 mm was applied. .............................. 913.27 Coronal (left) and sagittal (right) slices of PET images of rat 14 days after IV injection of Mn-52. This image was reconstructed without cascade correction and was not calibrated.Isotropic Gaussian smoothing with a FWHM of 1 mm was applied. .................................... 92xviii3.28 Transaxial (left), coronal (middle), and sagittal (right) slices through PET image of rat acquired from 75 to 90 min after IV injection of Mn-52. This image was reconstructed without cascade correction and was not calibrated. Isotropic Gaussian smoothing with a FWHM of 1 mm was applied. ............................................................................................... 923.29 Transaxial (left), coronal (middle), and sagittal (right) slices through PET image of rat 1 day after IV injection of Mn-52. Isotropic Gaussian smoothing with a FWHM of 1 mm wasapplied. ................................................................................................................................. 933.30 Transaxial (left), coronal (middle), and sagittal (right) slices through PET image of rat, 6 days after IV injection of Mn-52. This image was reconstructed without cascade correction and was not calibrated. Isotropic Gaussian smoothing with a FWHM of 1 mm was applied. .......................................................................................................................... 933.31 Coronal (left) and sagittal (right) slices through abdominal PET images of rat immediately after IV injection of F-18. ...................................................................................................... 953.32 Transaxial (left), coronal (middle), and sagittal (right) slices through head PET images of rat 75 min after IV injection of F-18 water. ........................................................................... 953.33 Sagittal slice of PET image of rat acquired 60 to 80 min after ICV injection of Mn-52.Image is centred on chest, and covers the posterior end of the brain and spinal cord extending back from it. Image is reconstructed without scatter or cascade corrections, and was smoothed to reduce noise. Image colour scale is shown at right as a rough guide. This scale, different from most other images shown in this section,  was chosen to better emphasize the curve of activity extending posterior from the brain (white blob at top centre), which appears to follow the curve of the rat spinal cord. An attenuation scanis shown in grayscale underneath the PET intensity for anatomical context. ....................... 973.34 Transaxial (left), coronal (middle), and sagittal (right) slices of PET image of rat head, acquired within hours of ICV injection of Mn-52. Approximate injection site is marked by the white arrow, and the olfactory bulb by the yellow arrow. .............................................. 983.35 Transaxial (left), coronal (middle), and sagittal (right) slices of PET image of rat head, acquired approximately one day after ICV injection of Mn-52. Approximate injection site is marked by the white arrow, and the colliculus by the yellow arrow. ................................ 983.36 Sagittal slices through PET images of rat brain 1 and 5 days after ICV injection of Mn-52 and additional non-radioactive MnCl2 targeted at the right lateral ventricle. ...................... 100xix3.37 Transaxial (left), coronal (centre), and sagittal (right) slices through PET images of rat brain 8 days after misplaced ICV injection of Mn-52 and additional non-radioactive MnCl2 targeted at the right lateral ventricle. ....................................................................... 1003.38 Transaxial (left), coronal (middle) and sagittal (right) slices of image of rat brain 1 day after IP injection of Mn-52 and 75 mM MnCl2 in approximately 2 ml volume. Isotropic Gaussian smoothing with a FWHM of 2 mm was applied. ................................................... 1033.39 Transaxial (left), coronal (middle) and sagittal (right) slices of image of rat brain 6 days after IP injection of Mn-52 and 75 mM MnCl2 in approximately 2 ml volume. Isotropic Gaussian smoothing with a FWHM of 1 mm was applied. ................................................... 1033.40 Transaxial (left), coronal (middle) and sagittal (right) slices of PET image of rat abdomen brain immediately after IP injection of Mn-52 and 75 mM MnCl2 in approximately 2 ml volume. Isotropic Gaussian smoothing with a FWHM of 1 mm was applied. ...................... 1043.41 Transaxial (left), coronal (middle) and sagittal (right) slices of PET image of ratabdomen 1 day after IP injection of Mn-52 and 75 mM MnCl2 in approximately 2 ml volume. PET emission data, with a red-white colour scale chosen for better visibility in this fused image than the rainbow scale used in other images in this section, are overlaid over attenuation map image (grey) for some anatomical context. The PET emission data were smoothed with an isotropic Gaussian smoothing with a FWHM of 1 mm. The transmission scan attenuation map data were smoothed with an isotropic Gaussian with a FWHM of 3 mm. .................................................................................................................1043.42 Sagittal slices of PET images of rat abdomen 1 day (left) and 4 days (right) after IP injection of Mn-52 and 75 mM MnCl2 in approximately 2 ml volume. The 1 day data was smoothed with a Gaussian with FWHM 2 mm, and the 4 day data was smoothed with a Gaussian with FWHM 1 mm. Scale of these images is deliberately compressed to emphasize a vertical (in the image) column of activity to the right (in the image) of the large out-of-scale region. This appears to be the spinal column, which is not visible at thewider scale seen in the previous figure. ............................................................................... 1054.1 Raw autoradiograph slice from brain of a rat that received ICV injection of Mn-52, cut perpendicular to the anterior-posterior axis. Colour scale ranges from 0 to 104 DLU (digital light units). ................................................................................................................ 111xx4.2 Raw autoradiographs of same slice of brain in a rat that received misplaced ICV injection of Mn-52 and non-radioactive Mn. At left, scale is 0 to 5x102 digital light units (DLU), and at right, scale is 0 to 5x104 DLU. ............................................................................................ 1124.3 Raw autoradiographs of the same slice of a rat that received a successful ICV injection of Mn-52 and non-radioactive Mn-52. The left and right images were exposed for 1 and 3 days, respectively. At left, the colour scale ranges from 0 to 500 DLU, while at right, the scale ranges from 0 to 1500 DLU. The structures visible in the images have similar coloursat these scales, but the longer exposure has better ability to see details of the activity distribution due to reduced noise. ....................................................................................... 1134.4 Raw autoradiographs of the same slice of brain in a rat that received an IP injection of Mn-52 and non-radioactive Mn. Both images have scale from 0 to 1x102 DLU. At left, the image was exposed for approximately 1 day starting 14 days post-injection. At right, the image was exposed for approximately 3 days starting 18 days post-injection. .................... 1144.5 Full raw autoradiograph of rat that received successful injection of Mn-52 and non-radioactive Mn into the right lateral ventricle. ..................................................................... 1164.6 Full raw autoradiograph of rat that received successful injection of Mn-52 and non-radioactive Mn into the right lateral ventricle. This image is rescaled to emphasize the appearance of the background distribution between the tissue slices that appear as dark red regions, off high end of the colour scale. ....................................................................... 1164.7 Autoradiographic image of slices of brain of rat that received misplaced ICV injection of Mn-52 and non-radioactive Mn. Blue rectangles indicate locations in which background intensities were averaged for each plate to estimate the constant background contributions. The left and right plates had separate constant backgrounds calculated. .............................................................................................................................1184.8 Autoradiographic image of slices of brain of rat that received successful ICV injection of Mn-52 and non-radioactive Mn, after subtracting constant background contribution. The intensities here are scaled to show remaining distribution of background between tissue slices. .................................................................................................................................... 118xxi4.9 Autoradiographic image of slices of brain of rat that received IP injection of Mn-52 and non-radioactive Mn, after subtracting constant background contribution. The intensities are scaled to show remaining distribution of background between tissue slices. This image is down-sampled more than the other rats' images due to fewer counts and resulting larger noise in the background, which makes it difficult to see the distribution without averaging pixels. ...................................................................................................... 1194.10 Single slice autoradiograph after background subtraction of rat that received IP injection of Mn-52 and non-radioactive Mn. Image has been down-sampled to reduce the appearance of statistical noise and better reveal structure in the activity distribution. .......................................................................................................................... 1204.11 Section of autoradiograph of rat that received ICV injection of Mn-52, centred between 4brain tissue slices (one at each corner). This image has been rescaled to show the spatially varying background after constant subtraction from the plate. Background outside of the tissue sections decreases with distance from the tissue, suggesting a similar pattern will continues within the tissue regions, making a simple background subtraction inadequate. ........................................................................................................1214.12 Estimated background distribution after convolving kernel with image intensity within tissue slice regions. Arbitrary scale. ...................................................................................... 1224.13 Estimated background distribution masked with tissue slice locations, and with a representation of the convolution kernel at centre. .............................................................1224.14 Full autoradiograph after background corrections have been applied. Colour scale is set to show remaining background variations after corrections. ............................................... 1244.15 Full autoradiograph after background corrections have been applied. Colour scale is set to show activity distributions within slices. .......................................................................... 1244.16 AR slices from brain of rat that received IP injection of Mn-52 that show large regions of damage. These slices were excluded from the volume for this animal, and were replaced with adjacent slices. ..............................................................................................................126xxii4.17 AR slice from brain of rat that received misplaced ICV injection of Mn-52 that shows minor damage and overlap of tissues. At the top of the slice, the tissue has a concave shape due to damage or lost tissue. At the bottom of the slice, another slice is partially overlapping. At the bottom-right, there is a streak of high activity at the edge of the slice,possibly due to overlap of the slice with itself. This level of damage was insufficient to exclude a slice from the tissue volume for this animal because most of the brain is unaffected, and the general shape of the distribution is intact. ...........................................1274.18 AR slices from brain of rat that received successful ICV injection of Mn-52. In the left image, a chunk of cortex is missing. In the right image, at the top and bottom, regions of increased brightness relative to the surrounding tissue, along with the flat border with the background region suggest that small section of tissue have folded over adjacent tissue, effectively combining their activities in the overlapping region and leaving the original location of the folded-over tissue as background. These levels of damage were insufficient to exclude slices from the brain volume for this animal. ................................... 1274.19 Autoradiography slice cut perpendicular to the superior-inferior axis after a successful injection of Mn-52 into the right lateral ventricle. The injection site appears as a distinct asymmetrical feature, marked by the white arrow. The colliculus appears as a symmetrical large accumulation, left of the black arrow, with the colliculus at its centre. The cerebellum appears with a moderate accumulation of activity, marked by the grey arrow. The cortex has a lower level of activity, surrounding the sides and front of the brain in the image. ................................................................................................................ 1294.20 Autoradiography slice cut perpendicular to the superior-inferior axis after a misplaced injection of Mn-52 into brain tissue near the right lateral ventricle. Activity appears in a V-shaped pattern near the brain mid-line, marked with the white arrow, possibly due to transport across the corpus collosum from the injection site. The colliculus appears, marked with the black arrow, as a region of higher uptake than the surrounding tissue. ... 1294.21 Autoradiography slice cut perpendicular to the superior-inferior axis after an IP injection of Mn-52. Activity appears in a similar pattern to the ICV injection, with the colliculus, marked with the black arrow, and cerebellum, marked with the grey arrow, brighter than the cortex, but without any bright spot in the lateral ventricles. ......................................... 130xxiii4.22 Autoradiography slices in a rat that lived for approximately three hours after ICV injectionof Mn-52. These slices was cut perpendicular to the anterior-posterior axis after injectionof Mn-52 into the right lateral ventricle. At this short time after injection, the distributionof activity in the brain is likely dominated by the positions and sizes of the intracerebral ventricles, through which Mn could spread without needing to cross a brain barrier. ........ 1304.23 Autoradiography slices in a rat that lived for approximately one day after ICV injection of Mn-52. These slices was cut perpendicular to the anterior-posterior axis after injection ofMn-52 into the right lateral ventricle. ...................................................................................1315.1 Precession of magnetization, M, in XY plane, perpendicular to external applied magnetic field, B0, that is aligned along the Z axis. Single turn signal reception coil is shown oriented with its surface parallel to the external applied field to detect oscillating signal from magnetization precession. ........................................................................................... 1365.2 Instantaneous precession of magnetization, M, about the effective applied magnetic field, Beff, that is the sum of the static main external applied field, B0, and the rotating RF field, B1. .................................................................................................................................1435.3 Excitation of magnetization, M, by rotation away from the direction of the main external applied field, B0. Shown in the rotating frame where the magnetization component in theX'Y' plane (the component perpendicular to the main applied field in the Z direction) does not precess about the Z axis. Excitation occurs due to precession of the magnetization about the RF field, B1, here shown oriented along the X' axis of the rotating frame. ...................................................................................................................... 1445.4 Exponential decay of transverse magnetization with two characteristic times. Different characteristic times produce different remaining magnetizations with time, leading to contrast in images of objects containing materials with different times. .............................1465.5 Exponential recovery of longitudinal magnetization with two characteristic times. Different characteristic times produce different magnetization recovery with time, leading to contrast in images of objects containing materials with different times. ............ 1475.6 Magnetization vectors in the transverse plane before (left) and after (right) application ofa 180 degree inversion pulse that rotated the magnetizations about the (vertical)y-axis. .................................................................................................................................... 148xxiv5.7 Magnetizations all oriented in the same direction of transverse plane immediately after excitation (left), all with the same phase. After some time precessing at slightly different rates (right), magnetizations have spread out in phase. Shown in a rotating reference frame. ................................................................................................................................... 1495.8 Magnetizations after phase spread during precession and inversion about the x-axis (left). After another period of precession, magnetizations have rephased to nearly the same direction, although some differences may remain due to irreversible contributions to precession rate variations. ................................................................................................1505.9 Relative magnitude of signal in steady state cyclical excitation of magnetization as a function of excitation angle. Signal curve is plotted for various TR/T1 ratios, as indicated by legend. ............................................................................................................................. 1536.1 In vitro saline R1 dependence on MnCl2 concentration at two temperatures. .....................1766.2 In vitro saline R1 dependence on temperature at various MnCl2 concentrations. ............... 1787.1 Rat in MRI bed, with snout in anesthesia delivery cone (at right), surface coil taped to head, signal transmission leads over its back, respiration monitoring probe attached to side, and rectal thermometer with lead in place (at left). .................................................... 1827.2 Look-Locker R1* fitting to data for 3 parameter fit. MR single-pixel signal intensity plotted against inversion delay (ms), with 3-parameter fit plotted as solid curve. ............. 1847.3 Illustration of search for solution to non-linear equation used to correct R1* measured byMRI to R1, as discussed in the Magnetic Resonance Imaging Background Flip-Angle Independence section. Corrected R1 values are shown on the x-axis, while cost function being optimized is shown on the y-axis. The corrected R1 value is found at the zero of thecost function. ........................................................................................................................ 1847.4 Midsagittal slices of T1 weighted MR images of rat head. Images were acquired at baseline, and after abdominal injection of MnCl2 at times indicated in the figure. In the 100 min image, ventricles (marked with white arrows) and pituitary (marked with yellow arrows) have pronounced localized signal enhancement. Pituitary enhancement increases at 1 day, and remains visible in the 1 week image, and may be present at 3 weeks. ................................................................................................................................... 186xxv7.5 Selected frames from sequence of T1-weighted images acquired as part of a Look-Locker inversion recovery T1-mapping acquisition. Data are acquired in a multi-slice imaging sequence, and reconstructed to produce images for multiple delay times after the inversion, as indicated in the figure. This acquisition was done at baseline, before the animal received any MnCl2 injections. .................................................................................. 1877.6 Midsagittal slice images of Mn concentration (scale in mM) in rat brain. Images are produced by subtracting R1 relaxation rate data from two separate image acquisitions: one at baseline, and one post-injection. Maps are shown for the times indicated in the figure after IP injection of MnCl2. Accumulation was particularly visible in the pituitary gland, outside the blood-brain barrier (marked with white arrows). ................................... 1887.7 Midsagittal slice of rat brain T1-weighted MRI acquired 1 day after ICV injection of MnCl2.Striking contrast enhancement is seen in the cerebellum (marked with arrow). ................. 1927.8 Coronal slice of rat brain T1-weighted MRI acquired 2 days after ICV injection of MnCl2. Structural details are seen throughout the brain. ................................................................ 1927.9 Coronal slices of (the same) rat brain T2-weighted MRI acquired at baseline (left) and 1 day after ICV injection of MnCl2. Unilateral darkening is seen near the site of injection (marked with arrow). ............................................................................................................ 1937.10 Midsagittal slices of (the same) rat brain R1 relaxation rate map. Scales in s-1. These maps were generated from data acquired before (top) and 1 day after (bottom) ICV injection ofMnCl2. Strong increase in R1 relaxation rate is seen in red and yellow regions of midbrain and spinal cord. Moderate increase in R1 relaxation rate in cyan regions elsewhere in brain. ..................................................................................................................................... 1947.11 Coronal slices of T1-weighted images of rat that received successful ICV injection ofMnCl2 and Mn-52 into the right (bottom in image) lateral ventricle. Images were acquiredat baseline (left), 1 day post-injection (centre) and 4 days post-injection (right). Strong signal enhancement is seen in the colliculus (marked with arrow). ..................................... 1967.12 Coronal slices of T2-weighted images of rat that received successful ICV injection of MnCl2 and Mn-52 into the right (bottom in image) lateral ventricle. Images were acquiredat baseline (left), 1 day post-injection (centre) and 4 days post-injection (right). ................196xxvi7.13 Coronal and sagittal slices of T1-weighted images of rat that received successful ICV injection of MnCl2 and Mn-52 into the right lateral ventricle, acquired 1 day after the injection. These slices are positioned to cut through the injection route (marked by arrows). .................................................................................................................................1977.14 Coronal and sagittal slices of T2-weighted images of rat that received successful ICV injection of MnCl2 and Mn-52 into the right lateral ventricle, acquired 1 day after the injection. These slices are positioned to cut through the injection route (marked by arrows). .................................................................................................................................1977.15 Coronal slices of R1 relaxation rate maps of (the same) rat which received ICV injection ofMnCl2 and Mn-52. Images were acquired at baseline and post-injection at times indicated in figure. Scale in s-1. .............................................................................................. 1987.16 Coronal slices of T1 weighted images at baseline (left), 1 day (centre), and 4 days (right) post-injection of rat that received misplaced right lateral ventricle ICV injection of MnCl2 and Mn-52. Reduced signal is seen in the area of the injected-side lateral ventricle (marked with arrows). .......................................................................................................... 1997.17 Coronal slices of T2 weighted images at baseline (left), 1 day (centre), and 8 days (right) post-injection of rat that received misplaced right lateral ventricle ICV injection of MnCl2 and Mn-52. Reduced signal is seen in a region between the lateral ventricles (marked with arrows). .........................................................................................................................1997.18 Coronal and sagittal slices of T1-weighted images of rat that received misplaced ICV injection of MnCl2 and Mn-52 targeted at the right lateral ventricle, acquired 1 day after the injection. These slices are positioned to cut through the injection route, which is marked by the white arrows. ................................................................................................ 2007.19 Coronal and sagittal slices of T2-weighted images of rat that received misplaced ICV injection of MnCl2 and Mn-52 targeted at the right lateral ventricle, acquired 1 day after the injection. These slices are positioned to cut through the injection route, which is marked by the white arrows. ................................................................................................ 2007.20 Coronal slices of R1 relaxation rate maps of (the same) rat which received misplaced ICV injection of MnCl2 and Mn-52. Images were acquired at baseline and post-injection at times indicated in figure. Scale in s-1. Large increases in R1 are seen localized near the siteof injection (marked with arrows). ....................................................................................... 201xxvii7.21 Midsagittal (left) and coronal (right) slices of T1 weighted images of rat brain 1 day post IP injection of MnCl2 and Mn-52. Signal enhancement is seen in the pituitary in the T1 weighted image (marked with yellow arrow), similar to figure 7.4. ..................................... 2027.22 Midsagittal slice of abdominal T1 weighted image of rat 6 days after IP injection of MnCl2 and Mn-52. The posterior brain (marked with white arrow) and spinal cord (marked with grey arrows) are visible. Motion artifacts are seen in the centre of the image, likely due to breathing. The large uniform region (marked by yellow arrow) may be the liver. ........... 2037.23 Coronal slice R1 maps of rat that received IP injection of MnCl2 and Mn-52 at baseline (upper left), 1 day (upper right), 4 days (lower left), and 6 days (lower right) post IP injection. Scale is in s-1. ......................................................................................................... 2047.24 R1 map coronal slices before (left) and after (right) resampling for coregistration. Loss of image sharpness and blurring is apparent in the resampled image. .................................... 2067.25 Relaxation rate change map between 4 days post ICV injection and baseline in rat that received successful right-later ventricle injection. Scale has units s-1. Largest increases areseen in the colliculus, and in the right lateral ventricle near the injection site.) .................. 2077.26 Relaxation rate change map between 8 days post ICV injection and baseline in rat that received misplaced right-later ventricle targeted injection. A distinct pattern with R1 change large near the injection site but less elsewhere is seen, along with decreases nearthe lateral ventricles. The scale (units of s-1) has negative values in this image, to illustratedecreased R1. ....................................................................................................................... 2077.27 Relaxation rate change map between 4 days post IP injection and baseline in rat. Scale has units s-1. The olfactory bulbs again appear bright, along with the midbrain and cerebellum. ........................................................................................................................... 2078.1 Screenshot captured during coregistration of Mn-52 autoradiograph (top row) to R1 difference image (bottom row) of rat brain using the ASIPro software Fusion tool. This tool allows images to be overlaid at varying opacity (centre row) to visually judge coregistration quality. Rotation and translations in three dimensions may be specified using the Fusion tool transformation interface (not shown). ............................................... 2118.2 Slices of R1 difference map (left, in units of s-1) and autoradiography image (right, in digital light units) of rat that received misplaced ICV injection of Mn-52 and MnCl2 after coregistration and applying a threshold filter. ...................................................................... 213xxviii8.3 Scatter plot of change in R1 between baseline and post-injection R1 maps in rat that received ICV injection of Mn-52 and MnCl2 into the right lateral ventricle against the autoradiography counts in corresponding pixels after coregistration of single slice brain images of both modalities. ................................................................................................... 2148.4 Scatter plots of change in R1 between baseline and post-injection R1 maps in rat that received misplaced ICV injection of Mn-52 and MnCl2 near the right lateral ventricle against the autoradiography counts in corresponding pixels after coregistration of single slice brain images of both modalities. These plots show the same data and fit, but the lower plot has a restricted scale to focus on the region with the majority of data points. .. 2158.5 Scatter plot of change in R1 between baseline and post-injection R1 maps in rat that received IP injection of Mn-52 and MnCl2 against the autoradiography counts in corresponding pixels after coregistration of a single slice brain image of both modalities. ............................................................................................................................ 2168.6 Coregistered single slices from images of a rat that received a right lateral ventricle injection of Mn-52 and non-radioactive MnCl2. At top left: PET image. At top right: AR image. Bottom right: Mn concentration image derived from MRI R1 relaxation rate change. Bottom left: same as bottom right, but smoothed with a 2 mm FWHM kernel to give the MR-derived concentration map the same resolution as the PET image. ................ 2198.7 Scatter plots in estimated concentrations of Mn or Mn-52 from MRI and PET in images ofthe same rat brain slice after coregistration between modalities. MRI concentration is calculated by applying a calibration factor of 4.2 s-1mM-1 to the change in R1 relaxation rate between baseline and post-injection images. At top, MR data is plotted unaltered. Atbottom, MR data has been smoothed with a 2 mm 3D Gaussian kernel to make its resolution similar to the PET image....................................................................................... 2208.8 Transaxial (left), coronal (centre), and sagittal (right) slices through PET image of rat head 1 day after IV injection of Mn-52 with high specific activity. Image has been smoothed with 2 mm FWHM kernel to reduce noise. ROIs covering the pituitary gland (?pit?) and brain (?brain?) are seen as yellow dashed lines in all slice  orientations. ............................. 224xxix8.9 Transaxial (left), coronal (centre), and sagittal (right) slices through PET image of rat head 1 day after IP injection of Mn-52 and non-radioactive MnCl2. Image has been smoothed with 2 mm FWHM kernel to reduce noise. ROIs covering the pituitary gland (?pit?) and brain (?brain?) are seen as yellow dashed lines in all slice orientations. .............................. 2248.10 Transaxial (left), coronal (middle) and sagittal (right) slices of PET image of rat abdomen brain 1 day after IV injection of Mn-52. PET emission data (red-white) are overlaid over attenuation map image (grey) for anatomical context. ........................................................ 2268.11 Transaxial (left), coronal (middle) and sagittal (right) slices of PET image of rat abdomen brain 1 day after IP injection of Mn-52 and 75 mM MnCl2 in approximately 2 ml volume. PET emission data (red-white) are overlaid over attenuation map image (grey) for some anatomical context. .............................................................................................................. 226xxxGlossary2D Two Dimensional3D Three DimensionalACC Animal Care CommitteeAR AutoradiographyARU Animal Resources UnitASL Arterial Spin LabellingBBB Blood-Brain BarrierBOLD Blood Oxygen Level DependentBP Binding PotentialBRC Brain Research CentreCSF Cerebral Spinal FluidCT (X-ray) Computed TomographyDAT Dopamine TransporterDLU Digital Light UnitsDTBZ DihydrotetrabenazineDV Volume of DistributionESR Electron Spin ResonanceEXFOR Experimental Nuclear Reaction DatabaseFBP Filtered Back ProjectionFDG FluorodeoxyglucoseFOV Field Of ViewFWHM Full Width at Half MaximumICV Intra-Cerebro-VentricularIP Intra-PeritonealIV Intra-VenousLOR Line Of ResponseLSO Lutetium OrthosilicateMDP Methylene DiphosphonatexxxiMEMRI Manganese-Enhanced Magnetic Resonance ImagingMIP Maximum Intensity ProjectionMIRD Medical Internal Radiation DoseMLEM Maximum Likelihood Expectation MaximizationMP MethylphenidateMR Magnetic ResonanceMRI (Nuclear) Magnetic Resonance ImagingNECR Noise Equivalent Count RateNEMA National Electrical Manufacturers AssociationNIST National Institute of Standards and TechnologyNMR Nuclear Magnetic ResonanceNNDC National Nuclear Data Center OSEM Ordered Subsets Expectation MaximizationPET Positron Emission TomographyPIB Pittsburgh Compound BPM PhotomultiplierPSTAR Stopping-Power and Range Tables for ProtonsR1 Longitudinal magnetization relaxation rateR2 Transverse magnetization relaxation rateRAC RacloprideRC Recovery CoefficientROI Region Of InterestRF Radio FrequencySOR Spill Over RatioSS Sum of SquaresSSS Single Scatter SimulationSTD Standard DeviationSUV Standardized Uptake ValueT1 Longitudinal magnetization relaxation characteristic timeT2 Transverse magnetization relaxation characteristic timeTAC Time Activity CurvexxxiiTE Echo TimeTI Inversion DelayTR Repetition TimeTOF Time Of FlightTx TransmissionUBC University of British ColumbiaUTE Ultra-short Echo TimeXCOM Photon Cross Sections DatabasexxxiiiAcknowledgementsAll those named in the preface are thanked for their contributions to this work. Without their advice and support, this project would not have been possible.Several other people assisted indirectly with this project, and are thanked as well. Stephan Blinder performed system maintenance for the PET scanner, and provided advice for software development during this work. Roxana Ralea prepared documentation for radiation shipments and provided assistance with access to and operation of the gamma spectrometer at TRIUMF. Stefan Reinsberg gave advice regarding MRI methods and provided software for reading MR images from his research group. Ryan Thompson provided computer systems administration support and data archiving assistance.xxxivMotivationThis work has three primary goals: to develop and characterize Mn-52 as a radionuclide for positron emission tomography (PET) imaging in phantoms and in small animals, to develop a quantitative magnetic resonance imaging (MRI) method for measuring Mn concentration in the brain of small animals, and to validate the MRI results by comparisons between modalities.The initial motivation for those goals arose from a combination of previous experience with PET and awareness of the technique of manganese-enhanced MRI (MEMRI). MEMRI uses Mn2+ as a contrast agent that is taken up by voltage-gated Ca channels. This leads to the accumulation of Mn2+ in areas of neuronal activation, allowing blood-flow-independent assessment of neuronal activity in the brainof small animals. PET is also an established method for measuring uptake of chemicals in the brain ofsmall animals. It was thus desired to develop an isotope of Mn for use as a PET radionuclide. Upon examining the decay properties of Mn isotopes, Mn-52 was selected as the most suitable radionuclide for imaging with PET.This idea was particularly appealing because MEMRI has established applications that could be replicated with PET, and MEMRI has several limitations where PET could perform better. In particular, MEMRI studies generally involve injections of MnCl2, often in large doses, which have potential toxic effects and may chemically or biologically alter the system being studied. PET can measure tracer concentrations of radioisotopes, which are orders of magnitude less than is given for MEMRI, and which may be better tolerated by animal subjects and may be scientifically more useful due to lack of toxic effects. As well, MRI is generally not a quantitative technique; most MRI methods, and most MEMRI studies in particular, do not attempt to measure concentrations of contrast agents or other basic physical properties of the object being imaged. PET can give quantitative results that accurately reproduce the concentration and distribution of the radiotracer in an object. This gives PET potentially more experimental power than MEMRI, in that variation in stimuli can be correlated with strength of response in terms of amounts or rates of tracer accumulation with PET.xxxvMRI also has advantages over PET, including better spatial resolution, and lack of a requirement to produce and handle radioactivity. The ability to localize Mn accumulation to specific MRI voxels and there measure absolute concentrations could allow experiments to measure localized stimulus-dependent accumulation. This could reveal the patterns of neuronal activation in response to stimuli, at a resolution better than possible with region-of-interest based methods, and with sensitivity to variations in amounts of accumulation within a single subject and between different stimuli. This could have substantially greater experimental power than existing MEMRI methods. It was thus desired to develop a per-voxel quantitative method to measure Mn concentration using MRI. Because the longitudinal magnetization relaxation rate, a parameter measurable with MRI, wasknown to be roughly linearly dependent on Mn concentration, and a recently-developed MRI pulse sequence for mapping the longitudinal relaxation rate was available and produced higher-quality images than was previously possible, a method based on this relaxation rate change related to Mn concentration was chosen.Although the longitudinal relaxation rate was expected to be roughly proportional to Mn concentration, it was anticipated that deviations from this proportionality would occur, of unknown severity and impact on quantification of results. It was thus desired to validate the results of MRI-derived concentration images by comparison with other modalities. Due to the relatively poor spatial resolution of PET, its use for validation of MRI in this work is limited to quantification tests after smoothing the MRI. It was thus decided to acquire autoradiography (AR) images of animals as well. AR has spatial resolution better than PET, and provides a useful standard for validation of MRI in this application. As well, because animals were already being prepared for imaging with Mn-52 PET, the same animals' brains could be prepared and imaged with autoradiography, and were thus a source of more data with little additional cost of resources or animal subjects.11 Positron Emission Tomography IntroductionPositron emission tomography (PET) is a medical imaging modality used in this work. This section contains background information regarding how PET data are acquired, how images are produced, various corrections and considerations necessary for quantitatively accurate results, and some of theanalysis that may be conducted on PET images.1.1 Source of PET ImagesPET images are generated by detecting 511 keV gamma rays produced by positron-electron annihilation. The positrons arise from the decay of positron-emitting radionuclides, including the commonly used F-18 and C-11, and less commonly used O-15, which are placed in the field of view (FOV) of a PET scanner. In this work, the radionuclide F-18 is used, as well as Mn-52, which is novel to PET. The process of positron emission during radioactive decay is fundamental to generating PET images.1.1.1 Radioactive DecayMany isotopes, including F-18 or Mn-52, are unstable and spontaneously decay into other energetically-preferred isotopes or states. Radioactive decay is a random quantum tunnelling process between a higher-energy initial state, and a lower energy state. On average, radioactive decay will cause an exponential decrease in the population of the radioactive isotope with time. Given the initial population N(0):N (t)=N (0)e??twhere the decay rate constant, ?, is a characteristic of an isotope.Because decay is a random process, when counting decays from a radioactive source in a fixed time, the standard deviation of the measurements will be roughly equal to the square root of the mean number of counts, ?. ?=? ?2This random nature of radioactive decays has consequences for PET image reconstruction, as discussed below. In place of the rate constant, the time taken for half of the initial population to decay, or half-life, of the isotope is often used to convey the timescale of decay of the isotope:N (T 1/2)N (0) =0.5=e??T 1 /2?ln (0.5)/ ?=T 1 /2Half-lives of different isotopes vary by many orders of magnitude: from nanoseconds to billions of years. For PET tracers, half-lives are typically on the order of minutes to hours. For example, C-11 hasa half life of 20.4 min, while F-18 has 109.8 min. The half-life of Mn-52 is somewhat longer than traditional PET tracers at 5.6 days.Radioactive decay processes include fission, alpha decay, pure photon emission, electron capture, proton emission, and both electron and positron emissions. What types of decay are possible and their relative likelihoods vary between isotopes, and in some cases, with the conditions and state of the unstable atom.The subatomic mechanism for radioactive decay involves the weak nuclear force and conversion between flavours of quark (for example, in the case of positron emission, an up quark changes into adown quark). The daughter nucleus itself may also be radioactive and undergo further decay processes, possibly rapidly enough to appear in practice as part of the same decay event.1.1.2 Positron DecayIn positron emission, the atomic number of the atom is reduced while the number of nucleons is conserved as a proton is converted into a neutron. During this process, other particles are emitted, including the positron (e+), an electron neutrino (?e), and possibly a number of gamma ray photons that depends on the parent and daughter isotopes and the particular decay path taken. The positron3will typically also annihilate with an electron (e-) in surrounding material, producing additional photons. Positron emission tends to occur for isotopes with relatively many protons and relatively few neutrons. It can only occur when the parent (initial, unstable) isotope and daughter (result of decay) have a difference in mass of at least two electron masses (2mec2 = 1.022 MeV). This excess mass is necessary to conserve mass-energy of the system, which includes the produced positron, as well as an ejected orbital electron due to difference of one in the number of protons in the neutral parent and daughter atoms.If the mass difference is greater than 1.022 MeV, positron emission is possible. The relative probability of decay by positron emission, or branching ratio, tends to increase with parent-daughtermass difference. For example, for decay of C-11 to B-11, the energy difference of parent and daughter nuclei is 1.9824 MeV (NNDC MIRD) and 99.8% of decays are by positron emission. For F-18 to O-18, the energy difference is 1.6555 MeV (NNDC MIRD) and only 97% of decays are by positron emission. However, with larger isotopes with higher energy difference, alternate decay pathways may exist, leaving a lower chance of positron emission. For decay of Mn-52 to Cr-52, the energy difference is 4.7114 MeV (NNDC MIRD), but only 29% of decays will include positron emission, while all decays include multiple gamma emissions.Regardless of the decay path, excess energy above 1.022 MeV between parent and daughter nuclei is conserved during the transition, and will contribute to energies of the decay products. For decay of F-18 by positron emission, the reaction may be written as:F918 ? O818 +e++?eThe presence of a neutrino in beta decay products allows simultaneous conservation of lepton number, energy, and momentum during the transition. The neutrino and beta particle share the excess energy of the transition in a continuous distribution. The positron thus has a range of possiblekinetic energies, from near zero, up to a limit determined by details of the decay path.4F-18, with its relatively simple decay scheme with no cascade gamma emissions, has a maximum positron energy determined by the full energy difference between parent and daughter isotopes: 1.6555 - 1.022 = 633.5 MeV. The average energy is somewhat less, due to sharing with the neutrino: 250 keV (Disselhorst, 2010) or 252 keV (Le Loirec, 2007).Mn-52 has a complicated decay scheme in which positron emission occurs by transition to an excitedstate of Cr-52 with a mass difference of 3.1138 MeV (NNDC MIRD) from stable Cr-52 and 4.7114 - 3.1138 = 1.5976 MeV from Mn-52 (NNDC MIRD). Subtracting 1.022 MeV from this energy gives the maximum positron energy of 575.6 keV. The remaining mass energy difference in this transition is dispersed to a variety of gamma ray photons emitted during the decay. The average energy of the positron is again somewhat less: 244.6 keV (Le Loirec, 2007).1.1.3 Positron AnnihilationAfter being produced in a radioactive decay from a PET tracer, positrons will typically annihilate with an electron in the surrounding material to produce two photons:e++e???+?These photons have energy approximately equal to the 511 keV rest mass of the annihilated beta particles, and are emitted in roughly opposite directions, preserving the net zero momentum in the interaction rest frame. Deviations from antiparallel orientation of photons of up to 4 mrad (0.2 degrees) occur in the majority of positron annihilations (Rickey, 1992), and a there is distribution of non-collinearity angles centred at 0 degrees with a FWHM of approximately 0.5 degrees in humans for F-18-FDG (Shibuya, 2007). As well, positrons are emitted with a range of kinetic energies dependent on the isotope, and will travel some distance before annihilation, as shown in table 1.1 for water, which is a good approximation of soft tissue.5Isotope Mean Range (mm) Maximum Range (mm)C-11 1.266 4.456F-18 0.661 2.633Mn-52 0.630 2.461Table 1.1: Theoretical ranges of travel in water before positron annihilation after emissionfrom PET radionuclides (Le Loirec, 2007).The positron range has an impact on PET image acquisition because the positions from which annihilation photons originate are displaced from the positions where the positron emissions occurred. This has the effect of limiting the resolution of the PET system, as the distribution being measured is in practice the locations of annihilations, rather than the distribution of radioactive atoms. Consequently, the measured data is smoothed or blurred by the random movements of the positrons before annihilation.Both displacement of the site of annihilation from the site of emission, and non-colinearity of the annihilation photons may contribute to incorrect estimation of the radioactivity distribution when the photons are measured by a PET system. 1.2 PET SystemA PET system is designed to detect positron annihilation gamma ray photons. When near-simultaneous photons are detected by the system in a suitable geometric configuration, the system records a coincidence event between those crystals. The recorded coincidence events are later reconstructed into an image of the spatial-temporal distribution of the radioactivity.A typical PET system consists of a cylindrical ring of detector elements, as seen in figure 1.1. The microPET Focus 120 system used for PET imaging in this work has a similar construction.6Figure 1.1: General Electric (Connecticut, USA) Advance human PET system with ring of detector blocks visible.Detector blocks and crystals may be distributed uniformly in the circumferential directions, or blocks of crystals may be arranged in a polygon around the field of view, possibly with gaps between adjacent blocks. Multiple rings of blocks of crystals may also be stacked in the scanner axis direction to extend the axial field of view of the system.1.2.1 ScintillatorsMost PET scanners detect gamma rays using rings of scintillator crystals. These crystals are composed of materials in which photons with energies between 100 keV and 1 MeV deposit energy primarily by Compton scattering or photoelectric absorption, generating electron-hole pairs. Typical scintillator crystals have a valence and conduction band gap of 5 eV or more if pure, but are doped with an ion that provides additional energy levels within the gap. After energy is deposited in the crystal by a gamma ray interaction, some of the doping ions are excited and relax, emitting visible or ultraviolet light photons, to which the crystal is transparent (Melcher, 2000). The number of visible photons released is roughly proportional to the energy deposited (Rooney, 1997), allowing the brightness of the visible light flash to be used to estimate the energy that was deposited by the gamma ray in the crystal during a gamma ray photon interaction.7Most PET systems produced in the last 10 years use lutetium orthosilicate (Lu2SiO5, LSO) crystals, including the microPET Focus 120 system used for PET experiments in this work (see page 56). LSO has a relatively short scintillation light decay time and a relatively high light yield, making it excellent for timing and energy measurements. It is also rugged and non-hygroscopic (Melcher, 2000), making manufacturing of detector modules less complicated than would be the case with some other crystals.1.2.2 PhotomultipliersScintillation photons in a PET scanner crystal are collected by photomultipliers (PMs) that are coupled to the crystals, producing electrical signals that are passed to attached electronics and eventually the data storage computer system. Individual crystals are typically grouped together in rectangular arrays, or blocks, sharing several PMs, such as illustrated in figure 1.2.Figure 1.2: Example arrangement of 7x7 grid of crystals (black) coupled to 4 photomultipliers (red).Actual PET systems will in general use different packing geometries than shown here for thephotomultipliers and relative sizes of the crystals and photomultipliers.Most PET systems used photomultiplier tubes (PMTs). PMTs have of a front plate that is transparent to the wavelengths of light produced by the scintillator crystals, over top of a photocathode that releases electrons by the photoelectric effect when struck by the scintillation photons. The tube has a potential difference along its length, so that the electrons are accelerated away from the plate. In the path of the electrons are a series of dynodes held at sequentially increasing voltages. As the electrons impact the dynodes, they induce secondary emission of additional lower-energy electrons.At each dynode, the number of electrons is multiplied, and after a series of dynode impacts, a 8greatly increased electrical pulse is generated when the electron cloud finally strikes an anode. The electrons striking the anode produce a pulse of charge, which is the output signal of the PMT. With properly tuned voltages, the total signal gain will be constant, and the output signal will be proportional to the number of photons initially striking the photocathode. This allows the output of the PMTs to give a measurement of the energy that was deposited in the scintillator crystal.More recently, solid-state PMs have become more commonly used in PET systems. One such device is the pixilated silicon PM, in which the scintillation light exposes a grid of avalanche photodiodes. Individual diodes may be triggered or not, at a rate roughly proportional to the brightness of the incoming light. The main advantages of silicon PMs are their small size in comparison with tube PMs,and their insensitivity to externally applied magnetic fields. Tube PMs involve charged particles travelling along their length, and exposure to a magnetic field will deflect the travelling electrons into curved paths, potentially moving them out of the intended path along the length of the tube. This can cause the tube PM to be nonfunctional. Silicon PMs can be designed to be unaffected by magnetic field exposure. This provides a means to integrate PET systems within magnetic resonance imaging scanners, which employ high magnetic fields in their design.1.2.3 Event CharacterizationBecause the electrical signal output from PMs is proportional to the brightness of the scintillation light, and thus the energy deposited in the crystal by the gamma interaction, the signal strength gives an indication of the energy of the gamma ray. Interactions by the photoelectric effect will deposit all energy in the crystal, while Compton scattering interactions may deposit less, and many such interactions lead to an event signal strength spectrum measured in the system. With an appropriate calibration factor between signal strength and deposited energy, the signal strength for each interaction will give an indication of that interaction's energy. Positron annihilation photon interactions will typically lead to a characteristic spectrum shape, with a prominent peak near the 511 keV due to photoelectric interaction depositing the full energy of the photon in the crystal, and a broad lower peak below due to Compton scattering in the detector crystal, which deposits only part of the photon energy. The spectrum may also contain different energies due to photons interacting in the crystal after Compton scattering interactions in the object being imaged, before 9reaching the detector crystals. Alternatively, some radionuclide decay paths produce gamma ray photons at energies other than 511 keV, which may also be detected by the PET system after the aforementioned interactions.To produce a PET image, it is generally desired to measure only pairs of simultaneous 511 keV gamma interactions in the crystals. In most imaging situations, these events most likely arise from positron annihilations, while photons of other energies are more likely to have other origins. The spatial localization of PET is based on the anti-parallel orientation that is characteristic of positron annihilation events, and the ability to identify a line of response (LOR) along which the annihilation occurred, between pairs of detector crystals that detect these photons, as illustrated in figure 1.3.Figure 1.3: Annihilation site (green), and paths of anti-parallel annihilation photons (blue), to the surroundingring of PET detectors (red). The PET system sees only which detector elements in which the photons aredetected but knows that the annihilation likely occurred in the field of view along the line connecting those elements.In order to selectively measure only simultaneous 511 keV positron annihilation gamma photons, timing and energy windows are built into the detector electronics. Whenever a signal is detected from the detector elements, the size of the signal is checked against a range, or window, that has been configured in the system. The allowed window of signal sizes is set to cover energies that are likely to be deposited in crystals by 511 keV gamma rays, so as to selectively detect positron annihilation photon interactions in the detector crystals, and selectively exclude other interactions. In this work, the energy windows used were generally 350-750 keV or 450-600 keV.10If the energy of a signal is within the allowed energy window, a timing mechanism in the detector system is primed to anticipate another signal within the energy window within a short period of time after the first. This time period, or timing window, was typically 5 ns for this work. If a second signal of acceptable energy occurs within the timing window of the first, the two signals are considered to be in coincidence, or effectively simultaneous, and can be referred to as a coincidence event. The timing window is used because positron annihilation photons are emitted simultaneously, and for most PET systems travel time is negligible, so coincidence events are likely tobe caused by an annihilation in the field of view and are recorded, particularly when the time between individual radioactive decays is large in comparison with the timing window.The choice of energy and timing windows is non-trivial. For scintillator crystals with slow rise and decay times, the timing of individual photon interaction signals may be imprecise, and require a correspondingly long timing window. A longer window is not desirable, however, as it increases the chances of independent decay events producing photons that are detected within the timing window, leading to a random coincidence that lacks the essential spatial information for PET imaging. Too short a window may exclude pairs of detected photon interactions that did arise from asingle annihilation, but which had a relatively long difference in the times when they were detected due to variations in measurement hardware.Similarly, energy windows that are very tightly constrained around 511 keV will be very selective for positron annihilations, reducing the chance that photons of other origins will be counted. However, a very tight energy window will exclude many positron annihilation photons that interacted by Compton scattering instead of photoelectric absorption, and even many unscattered photons when the window is small relative to the system energy resolution. The ideal energy window for any given experiment will thus depend on the particular imaging situation, with factors such as the amount of scattering, the presence of external sources of photons or other non-annihilation photons in the vicinity of the PET detectors, and possibly other practical concerns such as whether the system is routinely used at a particular energy or timing window and thus has normalization and calibration information already available.111.3 ReconstructionPET image reconstruction can be viewed as an inversion problem. PET data includes series of measured coincidence events, which include a time when the event was measured, and spatial information in the form of which detector crystals were involved. The goal of reconstruction is to recreate the three- or four- dimensional spatio-temporal distribution of radioactivity within the field of view of the PET scanner during acquisition. The measured data are a transformation of that distribution. That transformation can be modelled in various ways, and various algorithms can be employed to invert it (or estimate its inversion).1.3.1 Forward ProjectionTo discuss reconstruction algorithms used to invert the transformation of the PET measuring process, it is useful to first discuss how the measurement process may be modelled as a forward transformation. Forward transformation, or projection, models how the distribution of radioactivity in the field of view leads to the events and data measured by the PET scanner.PET data are generally recorded as a list of coincident photons detections in pairs of detector elements. The number of these detections may be modelled as a geometrical projection of the 3D radioactivity distribution in a volume around a theoretical line connecting those detector elements. This line, referred to as a line of response (LOR), is used to represent the volume within the scanner field of view where positron annihilations may produce pairs of gamma ray photons that could travelto and interact with that pair of crystals.Sets of LORs oriented with the same angle but different radial offset may be combined to form a 1-dimensional projection of, as illustrated in figure 1.4, parametrized by radial offset, r, at various angles ? around the field of view of a PET scanner.12Figure 1.4: Example single plane cut across the field of view of a PET scanner (enclosedby red circles) containing a source distribution (solid black), and a single angularprojection (at right) of that source distribution, as measured by the PET system.1.3.2 SinogramsA common first step in PET image reconstruction is to histogram the list mode data into projection bins corresponding to lines with evenly-spaced angles and radial offsets from the scanner axis through the field of view. The sinogram consists of an array of numbers for each bin, each indicating the number of coincidence events that were histogrammed into that bin. Generally there will be more pairs of detector elements than sinogram bins, so a bin will have coincidence events sorted into it that were recorded from multiple pairs of detector elements.The term 'sinogram' arises from the appearance of a point source of intensity in an image volume that has been projected and displayed as a sinogram. If displayed as a 2D image, a point source that is offset from the centre of rotation of the sinogram will appear as a sine-wave shaped curve through the image, as seen in figure 1.5. In practice however, most PET images do not exhibit this shape clearly due to their more complicated overlapping projections from a distribution of intensity or radioactivity in an image or source volume.13Figure 1.5: Single slices of sinogram data. Slices images both have horizontal axes in theradial direction. The left image has the vertical axis in the circumferential direction, and showsthe characteristic sine-wave shape from which sinogram get their name. The right image has verticalaxis in the scanner axial direction, and shows an axial series of projections from the same angle.Histogramming is useful because it simplifies and reduces the raw list mode coincidence data into a more manageable form. The number of sinogram bins is often substantially less than the number of LORs in a scanner, or the number of coincidence events recorded during a scan. As an approximate example, with 13.8x103 detector elements in a PET system (such as the microPET Focus 120), each ofwhich can record coincidence events with any of the other crystals, a coincidence event could be detected in any of approximately (13.8x103)2 = 1.9x108 different lines of response. If a PET scan records 108 coincidence events (equivalent to 2.8x104 coincidences per second for a 1 hour scan), the average number of coincidences per line of response is less than 1. Alternatively, a sinogram file with 128 radial offsets and 144 circumferential angles per sinogram, and 1567 sinograms at differentaxial position and plane angles (as is the case for the microPET Focus 120, see page 56), has a total of 2.9x107 sinogram bins. A sinogram file may be further reduced by rebinning data acquired from allplane angles into transaxial sinograms that are perpendicular to the scanner axis. This rebinning is commonly done using a technique referred to as Fourier rebinning, which estimates the spatial frequency content of the transaxial sinogram planes from that of the measured tilted planes (Matej, 1998).With these numbers of bins, LORs, and events, the number of and simplicity of geometrical calculations with sinogram-binned data is less than with the data in list mode with unhistogrammed lines of response. 14Additionally, correction factors are needed during PET reconstruction, including attenuation and normalization factors (see page 17). These factors are calculated for each sinogram bin, separately from the emission data acquisition, and in the case of normalization, are reused for multiple image acquisition and reconstructions. It would require substantially more memory and time to calculate attenuation and normalization factors for each LOR, rather than for each sinogram bin.1.3.3 Filtered Back ProjectionAn important method for reconstructing PET data is filtered back-projection (FBP). FBP takes measured PET data, which are essentially projections of a radioactivity distribution in the scanner onto 2D planes, and reconstructs an estimate, or image, of the 3D activity distribution. In FBP, the activity in each sinogram bin or LOR is "back projected" by assigning values to image points through which the LOR (or LORs corresponding to the bin) passes. These are locations where positron annihilation gammas on the LOR may have originated. Traditional PET (i.e. without time of flight information) provides no information about where along that line an annihilation occurred. FBP algorithms thus have no means to select one point over another along a projection line, and instead assign events from a bin to all points along that bin's line through the reconstructed image. The combination of full radial sampling of sinogram bins and sufficient radial offsets provides sufficient information to reconstruct a 3D volume by this means. Simple backprojection is insufficient however, as the process adds a blurring to the image which is mathematically described in position space as "1/r". That is, the blur is similar to that caused by convolution with a kernel that is inversely proportional to radial distance from point in the image. The underlying cause of this blur lies in the combination of forward projection as part of acquiring the PET data, and simple back projection, which is not an inverse of forward projection. The presence of this blur is the reason why filtering is included in the back projection process used for PET image reconstruction; suitable filters such as a truncated spatial-frequency ramp filter will reduce or, in the ideal case, remove this blur.15A notable consideration with FBP reconstruction is the appearance of noise in the measured data and in reconstructed images. PET systems measure the products of radioactive decay, which, as noted above, is a random process, and will inevitably result in noise in measured data. As well, real objects such as patients, animals, or phantom objects generally have distributions of radioactivity that are relatively sparse in higher spatial frequencies. Consequently, the ratio of noise to signal in higher spatial frequencies tends to be higher than in lower spatial frequencies. A filter that amplifies the higher spatial frequencies of projection data will thus tend to preferentially amplify noise, rather than meaningful signal information.Another effect of noise in FBP reconstruction is inconsistency of projections. As noted above, the individual projection measurements are independent realizations of random counting experiments. Due to the random variation of the results of such counting, it is inevitable that different projection components will have somewhat inconsistent measured values. Inconsistent, in this context, indicates that the projections measured do not represent the same underlying activity distribution. If this is the case, it may lead to artifacts in reconstructed images, such as a streak across an image corresponding to a sinogram bin containing a randomly high number of counted coincidences which are not reflected in other bins that pass through the same regions of the source distribution.Even without random variation in projections, the limited number of sampling directions in a real PET system has the potential to break the approximation of continuous sampling as used above. Particularly for small, isolated sources of activity, radial streak artifacts may appear in PET images, as illustrated in figure 1.6.Figure 1.6: Slice of PET images of 1.1 mm inner diameter capillary tube containing PET tracer.Image scale was adjusted to emphasize the appearance of radial streak artifacts in FBP reconstruction.16A notable advantage of FBP is that images it produces are linearly dependent on the input projectiondata. There is no noise-related bias in the images, and data may be measured data may be summed before or after reconstruction and will produce the same image.Filtered back projection is fast and can be very effective, but can also be limited. It is necessary that the measured data have uniform angular sampling, which is not possible with some PET scanners. Aswell, the filtered backprojection algorithm has limited ability to properly compensate or correct for physically understood effects of the measurement process, including positron range or annihilation photon non-collinearity, or statistical variation in measured results. The PET system used in this workwas the microPET Focus 120, which does provide uniform angular sampling, and filtered backprojection was used to reconstruct PET images.1.3.4 Other ReconstructionsWhile filtered back projection (FBP) is used exclusively in this work for PET image reconstruction, there are alternative reconstruction methods for PET. A notable class of these are the iterative reconstruction algorithms, which are particularly useful when it is not possible to use a back-projection-based reconstruction algorithm, such as when angular sampling of data is not uniform. Aswell, there are numerous physical corrections that cannot be modelled easily within a back-projection framework, but which can be incorporated into iterative methods.Iterative reconstruction algorithms operate by taking an initial estimate of the image that representsthe activity distribution that was measured, and forward-projecting that image to generate estimated projection data. The projection data is in the same form as is measured by the system, so can be directly compared with that measured data. From the differences between what is forward projected and what was measured, the image is corrected, and the projections generated again. Thiscycle repeats until the desired level of agreement between the measured and estimated data is reached.171.4 CorrectionsThe model of PET data acquisition involves detection of pairs of antiparallel 511 keV positron annihilation gamma ray photons. Various physical and practical considerations complicate and pollute the measurements, however, and must be accounted for when reconstructing measured data into images.1.4.1 RandomsPET scanners record random coincidences when 511-keV gamma rays from two independent radioactive decays are detected within the coincidence window. Because the gamma rays arise from independent decays, they are generally not collinear, and do not convey the spatial information of coincidence events arsing from a single positron annihilation. These must be corrected for during reconstruction to ensure images accurately represent the distribution of activity in the field of view. Positron emissions from radioactive decay occur with random timing, and each decay occurs statistically independently of all other potential decays. For a random coincidence to occur, two suchrandom decays must occur nearly simultaneously, within the timing window of the PET system. For small measuring periods, ?t, the rate of random decays may be writtendP (2)d?t ?(n ?)2 ?twhere n? is the rate of decays.Correcting for random coincidences requires having an estimate of their distribution across the measured LORs in a PET scanner.One method for arriving at such an estimate is to acquire coincidence data after introducing a delay into the coincidence timing window which is considerably longer than the duration of the window. Rather than recording photon interactions that occur effectively simultaneously, and thus likely 18arose from a single positron annihilation, the system instead records a pair of photon interactions that occurred well separated in time, and thus could not have arisen from the same annihilation. In this manner, the system measures only events that have none of the spatial information provided by true coincidence events due to antiparallel positron annihilation photons. Events recorded with sucha delayed timing window will instead have the same distribution as random events that occur due tonear-simultaneous detection of photons arising from separate positron annihilations. As well, as long as the timing window is of the same duration as that used during the main data acquisition, andthe amount of activity in the field of view is effectively equal, the rate of random coincidences is expected to be the same in both acquisitions. This method is used by the Focus 120 microPET system used in this work.Alternatively, the distribution of random coincidences may be estimated from the rates of single events detected by the PET system. Single events occur when the measurement system detects a single photon interaction, within the acceptable energy window,  but does not detect a second such interaction within the timing window for accepting a coincidence.The rate of random coincidences detected on a line of response between two detectors with singles rates S1 and S2 with a timing window ?t is the probability of measuring an event on the second after an event on the first, multiplied by the rate of events on the first:R12?S 1S 2 ?tUsing this relationship, or the delayed coincidence window method, the randoms distribution for measured activity may be estimated. With this distribution, the measured coincidences may be corrected by subtracting the randoms prior to reconstruction, or the randoms may be added as a term in an iterative reconstruction forward projection step as discussed above.1.4.2 AttenuationPET scanners record positron annihilation gamma ray photon interactions in the ring of detectors surrounding the field of view. To be measured, annihilation photons that originate within an object must travel from the site of positron annihilation, through some portion of the object, and on to the 19detectors. While passing through the object, the photons have some chance to interact with the object's mass, and may be absorbed or scattered out of their initial path. In either case, that photon may not reach the detector crystal along its original path, and a potential photon interaction and coincidence with another photon interaction from the same annihilation will be lost to the system. This loss of counts is referred to as attenuation. The impact of this attenuation on PET imaging is thatactivity sources placed near or within a mass will be partly obscured by that mass, and the number of annihilation gamma ray photons from the source that the detector measures will be reduced. Thiseffect must be corrected in order for a PET scan to produce an accurate estimate of the activity in the scanner.As noted earlier, the spatial localization of PET is based on the anti-parallel orientation that is characteristic of positron annihilation events, and the ability to identify a line of response (LOR) along which the annihilation occurred, between pairs of detector crystals that detect these photons. This ability to constrain the path of detected photons to a single line through the object is also important for attenuation correction, as the amount of attenuation along an LOR will depend on the attenuating properties of the object for 511 keV gamma rays along that LOR.Photon attenuation in materials is modelled similarly to radioactive decay in time; each small segment of material through which a photon passes has some chance of interacting with the photonthrough several possible mechanisms. The most important such mechanisms for attenuation of energies of photons in PET are incoherent Compton scattering, which is dominant in most biological tissue, and the photoelectric effect, which becomes important in high atomic number materials suchas lead. Coherent scatter may also have a relatively small contribution.The amount or rate of attenuation in an object may be described by linear attenuation factors, which assume a uniform volume of material of a fixed density, and indicate the attenuation with distance travelled. In most soft biological tissues, the attenuation factors will be similar to water. Bone and volumes of air within soft tissue are frequently treated as distinct tissue types, with substantially different attenuation factors, as seen in table 1.2.20Material ? (cm-1)Air 0Water 0.095Bone 0.178Aluminium 0.220Table 1.2: 511 keV photon linear attenuation factors (?) for selected materials from the microPET Focus reconstructionsoftware. These factors account for all interaction mechanisms that contribute to photon attenuation.The effect of attenuation on photons travelling through tissue may be described mathematically using the above attenuation factors.  The total amount of material through which the two annihilation photons must pass for both to reach PET detectors is the same regardless of where along the line they are generated, and their probabilities of interaction are independent. The probability of two photons originating on and oriented in opposite directions along a line connectingtwo detectors to both reach those detectors is determined by the total amount of attenuating material along that line.Pboth reach=e?r '=r Det1r Det2?(r ' )dr 'Correcting measured PET data for attenuation effects generally involves applying multiplicative scaling correction factors to the number of coincidences measured along LORs or in sinogram bins, to compensate for counts lost from those measurements due to attenuation. These correction factors may be estimated by assigning attenuation values to regions of an image volume, and then integrating the factors along paths corresponding to sinogram bins or LORs. These values may be assigned based on prior information, possibly guided by an initial reconstruction without attenuation correction, or using other sources of geometrical information and tissue classification such as an anatomical magnetic resonance imaging (MRI).Alternatively, attenuation factors may be determined using transmission measurements with x-ray (up to roughly 100 keV, arising from atomic electron state transitions) or gamma ray (above roughly 100 keV, arising from nuclear state transitions) photons. These measurements may be done with computed tomography (CT) equipment, which is often incorporated into clinical PET/CT scanners, or 21by measuring the transmission of photons from a moving source using the PET scanner's own photon detection hardware.The microPET Focus120 used in this work has a rotating point source holder for acquiring transmission scans to determine attenuation factors of objects being imaged. In this work, a Co-57 point source was used for transmission scans. The measured attenuation factors are used for correcting the number of coincidence events measured on each bin or line of response during reconstruction. The effect of attenuation correction on a reconstructed image is shown in figure 1.7.Figure 1.7: Profiles through image of uniform phantom of 7 cm diameter containing F-18 reconstructedwith and without attenuation correction, with profile values scaled to be similar at edges whereattenuation effects are minimal. The uncorrected image has distinctly lower values near the middle, whereactivity is surrounded by the most attenuating material.1.4.3 Scatter511 keV gamma rays travelling towards a detector crystal may be scattered out of their path by the material in the field of view. The majority of photon interactions in biological tissue will be Compton or incoherent scatter, in which the interacting photon is redirected along a different path from its original trajectory. This may result in the photon being counted in a different crystal from its original 22destination. Depending on the energy of the scattered photon and the energy window of the system, the scattered photon interaction may still be measured in coincidence with its annihilation partner. Because one (or potentially both) gammas were scattered away from their original path, theline of response between the crystals where the gammas were detected will likely not pass through the location where the positron annihilation occurred. This leads to a background of scattered events with a distinctive pattern, different from un-scattered coincidences or random events. As with random events, scattered events often must be corrected in order for reliable PET images to be reconstructed.The distribution of scattering angles produced by incoherent or Compton scattering is described by the Klein-Nishina equation, which is derived from quantum electrodynamics:d?d?=?2r c2P (E ,? )2 [P (E ,? )+P (E ,? )?1?1?cos2? ]/2whereP (E ,? )= 11+( Eme c2)(1?cos?)and rc = ?/mec, me is the mass of an electron (511 keV/c2), ? is the fine structure constant that characterizes the strength of photon-electron interactions, ? is the scattering angle of the photon, E is the initial energy of the photon being scattered, and d?/d? is the differential cross section for scattering at a particular angle. The total Compton scattering cross section ? can be determined by integration over all solid angles.The energy of the photon after scattering E' may also be described:E ' (E ,?)=E P (E ,?)At 511 keV incident photon energy, the angular distribution of scattering angles is peaked in the forward direction. As well, for higher angles of scattering, there is a larger reduction in the energy of the scattered photon, giving a tendency for higher angle scattering to be excluded from PET data dueto the energy window of the system. As such, measured scattered annihilation photons tend to have 23been scattered in a direction close to their original trajectory, but with an angular offset sufficient to make the photon interaction useless for determining the location of the positron annihilation.Unlike random coincidences, there is no method to directly measure the distribution of scattered coincidences in measured PET data for arbitrary objects. In order to implement scatter correction, it is thus necessary to use a model of the scattering in an object to estimate the scatter distribution. Most such models will make use of the Klien-Nishina formula to determine the angular distribution of scattered events, in addition to an attenuation map of the object to determine the likelihood of scattering interactions at various locations in the object.After it is generated, the estimated scatter distribution may be rescaled to match the amount of scatter present in the coincidence data, which may include contributions from activity outside the field of view of the scanner. This rescaling may be done using a subset of the sinogram bins in which there is expected to be no un-scattered non-random coincidences. Any bin or line of response that does not pass through the object being scanned cannot measure coincidences that arise from a single annihilation without scattering, so these bins will contain only randoms and scattered events (or cascade events, as discussed below). If randoms have already been corrected for, any activity in these bins may be attributed to scattered events, and used for determining the scaling of a scatter distribution during scatter correction. After rescaling, the scatter distribution is subtracted from the measured sinogram data, which may then be reconstructed as a scatter-free image.Iterative reconstructions may instead incorporate a scattering estimate when forward-projecting image data to projection data, using the measured attenuation information to determine the amount of scatter that is expected for the current image estimate.1.4.4 Cascade CoincidencesIn addition to scattering and random coincidences, another source of coincidences that do not match the desired distribution for PET reconstruction are the cascade gamma ray emissions in the decay of non-pure positron emitting radionuclides. Particularly relevant for this work are the cascadegamma emissions of Mn-52. Nineteen different gamma emissions energies are produced from Mn-2452 decay, as shown in figure 2.1, although not all will occur during a single decay. During decay pathsthat include a positron emission (which are 29% of Mn-52 decays), two or three gammas may occur with high probability that have energies making them likely to be detected and affect the measured coincidences data, which are shown in table 1.3.Emission Probability per Decay(%)Gamma Energy (keV)2 x Annihilation 29.3 511.0Gamma 90.0 744.2Gamma 94.5 935.5Gamma 100 1434Table 1.3: High probability Mn-52 decay cascade gamma emissions (NNDC Database).Cascade gamma interactions may arise from the same decay event as 511 keV annihilation photons that the PET system is trying to measure. Consequently, like scattered gammas, cascade gammas may be detected by the PET system in coincidence with those annihilation photons, at a rate that is aconstant fraction of coincidences involving only annihilation photons. This is in contrast to random events, for which the fractional rate of coincidences varies with the total activity in the scanner.Several algorithms have been developed for cascade correction. The general theme of most is similarto the scatter and randoms distributions discussed above, in which a distribution is estimated, whichis then fit to the radially peripheral components of measured data. Estimation of the distribution of cascade events may be done by several means. Like random events, cascade gamma coincidences convey very little information about the distribution of activity in the scanner. The distribution is distinct from the scattered or randoms distributions (LaForest, 2009), but can be approximated by the randoms distribution (Watson, 2008), or even a constant distribution across sinogram bins (LaForest, 2009). The cascade distribution may also be estimated by convolution of a kernel with the measured emission data (Beattie, 2003).25The rigorous calculation method (LaForest, 2009) for estimating the cascade distribution was not used in this work (see page 62), so will not be discussed in detail. Briefly, given an attenuation map for an object, the total attenuation for each activity source voxel to each detector element may be calculated. Given an initial activity reconstruction, without cascade correction, and the attenuation for each voxel and detector, the distribution of cascade events arising from the activity distribution emitting angularly uncorrelated pairs of photons may be estimated, by calculation or simulation.Approximation of the cascade distribution as similar to the randoms distribution is particularly usefulwhen the object being imaged is small (under 10 cm diameter) and has relatively little attenuation that might affect the cascade photons differently from collinear photons originating at the same location, and when the cascade photons have higher energy and thus less attenuation. Objects positioned near the centre of the scanner field of view are also more suitable for this approximation (Laforest, 2009). Imaging with small animals generally meets these criteria, suggesting the randoms approximation will be useful in this case (see also page 66). Additionally, it is necessary that there belines of response that measure coincidences arising predominantly from cascade gamma or random coincidences, and that have few or nor scattered photons or unscattered annihilation photons. These LORs are necessary to provide a target level for rescaling that accurately represents the amount of randoms-shaped background in the measured data.The rescaled-randoms subtraction method is also relatively simple, and was thus chosen for use in this work for cascade correction. The details of implementation and results of cascade correction in this work are discussed in the PET imaging section (see page 62).1.4.5 NormalizationAn ideal PET system would be uniformly likely to detect potential interactions of gamma rays in any of its detector crystals, and to record coincidences between any two crystals through which gamma rays pass. In practice however, variations in the crystals and electronics lead to differences in sensitivities for gamma interactions. This presents a problem for PET reconstruction, as variations in this sensitivity between crystals will appear in measured data, and will produce variations - most prominently streaking - in reconstructed images that are unrelated to the actual activity distribution 26in the field of view.To correct for this effect, normalization factors generated to compensate for the variations in the crystal sensitivities. These factors are generated by acquiring data on the scanner to measure crystal sensitivities. There are several methods to do this, and associated means for converting measured data to normalization factors.One such method is to acquire an image of a uniform phantom object containing radiotracer. The cylindrical phantom is placed symmetrically in the centre of the field of view, with its activity volumeintersecting most of the lines of response typically needed for imaging experiments. For this symmetrical system of known size, the expected activity ratios between various lines of response or sinogram bins that pass through the object are calculable. Any anomalous variation in the expected to actual measured data may be attributed to normalization effects, and the differences may be used to generate normalization factors for correcting subsequent non-uniform object imaging data.Alternatively, a rotating point source may be used in place of a uniform cylinder of activity.Direct normalization of each line of response in the above manners is often impractical, however, in that they will not properly normalize lines of response that do not pass through an object, or because they require a very high number of counted coincidence events for statistically adequate data for accurate results. In practice, so-called component- or model-based normalizations are used instead. These methods employ knowledge of symmetries of the detector system and source distribution to reduce the variance in correction factors.Single sinogram slices of normalization factors, un-normalized emission data, and normalized emission data are shown in figures 1.8 through 1.10.27Figure 1.8: Normalization factors sinogram. Measured data sinograms are rescaled bin-by-binduring reconstruction to compensate for these normalization factors.Figure 1.9: Raw sinogram data before normalization factors are applied.28Figure 1.10: Sinogram data after normalization factors are applied,with substantially reduced pattern of streaks.1.4.6 CalibrationThe previously-mentioned correction factors are generally necessary to ensure that a reconstructed PET image accurately represents the distribution, or shape, of the radionuclide in the field of view. They do not, however, provide enough information to accurately reproduce the precise concentration or amount of activity that was imaged. The precise number of coincidences that a PETsystem will measure from a given amount of activity depends on factors that cannot be derived fromthe measured data, and which are prohibitively difficult to calculate from known or measurable properties of the materials and electronics in the system.Instead, PET calibration is performed by imaging a phantom object containing a known concentration of radiotracer, reconstructing an uncalibrated image, and determining the appropriatecalibration factor to match the image activity to the known phantom activity. This calibration factor is then used to update the scaling of the normalization data used in reconstruction of the uncalibrated image. Future reconstructions using the now-calibrated normalization data will then have output intensities that represent their actual radiotracer concentrations.291.5 PET ExperimentsConducting a PET imaging experiment involves production of a molecule that includes a radioactive atom, or radiotracer, introducing the tracer into a biological system or phantom object, acquiring data, reconstructing, and analyzing the resulting images to determine biological or physical properties of interest that motivated the experiment. Details of each of these steps will vary greatly between different imaging experiments. A common framework is to calculate the amount of activity within a region of the image representing a structure of interest, and to compare this amount with the injected activity. Alternatively, the time activity curve (TAC) in a region of interest (ROI) may be fit with a kinetic model that describes the exchange of the radiotracer between different tissue compartments, parametrized by exchange rate constants or other parameters of biological interest.This section gives examples of radiotracers used in PET, and briefly describes analysis of PET images after reconstruction.1.5.1 Biology and TracersThere are an innumerable variety of PET tracers in use and that can or could be produced. A few examples covering a variety of applications are discussed below, including the most commonly-used PET tracer in clinics, F-18-fluorodexoyglucose (FDG), F-18-fluoride, several dopaminergic tracers usedfor Parkinson's disease imaging research, and Pittsburgh compound B, which is used for Alzheimer's disease imaging.FDG consists of an F-18 atom bound to a molecule that closely mimics the properties of glucose in vivo (Vallabhajosula, 2007). The tracer is injected into a subject, and will accumulate throughout the body, most notably in regions of high metabolic activity. This accumulation occurs because FDG is brought into cells that are metabolically active, much like glucose, where it is broken down, leaving the F-18 within the cell. This is of particular use for cancer imaging, as masses of rapidly dividing cells will consume disproportionately high amounts of glucose, leading to localized accumulation of F-18 in tumour regions. This allows PET scans to locate tumours in vivo. FDG is also used in cardiac imaging to assess myocardial metabolic activity, which is useful for treatment planning with coronary30artery disease (Allman, 2010).Another F-18 based tracer is F-18-fluoride, which is used to characterized bone disease and monitor response to therapy. After injection, F-18-fluoride is bound at the surface of bone crystals, particularly at sites of bone remodelling with high turnover (Even-Sapir, 2007).Another class of PET tracers are the dopaminergic tracers used for preclinical studies of Parkinson's disease (PD). These tracers include the C-11 labelled dihydrotetrabenazine (DTBZ), methylphenidate (MP), and raclopride (RAC). All three bind selectively to receptor sites in the brain that are involved in the dopaminergic signalling pathway, particularly in the striatum. DTBZ binds most strongly to the vesicular monoamine transporter type-2, located in the presynaptic nerve terminals projecting into the striatum that deliver dopamine. MP, used clinically with the commercial name Ritalin, binds to the dopamine transporters (DAT) which reabsorb dopamine from the synapse into the presynaptic neuron. RAC binds most strongly to the D2 dopamine receptors in the postsynaptic neuron. These tracers, used in combination, can reveal extensive details of the health or function of the dopaminergic system in human patients or animal models.Another notable PET tracer is C-11 Pittsburgh compound B (C-11-PIB; Rabinovinci, 2007). PIB binds to amyloid-beta plaques that form in the brains of Alzheimer's disease. This is useful for discrimination between Alzheimer's and other dementia such as frontotemporal lobar degeneration.1.5.2 PET Image AnalysisPET images may be reconstructed into volumetric concentration maps. The values of the voxels in such an image give estimates of the average concentration of radiotracer in a corresponding volume in the scanner during acquisition. Concentration maps may be themselves of interest, as was the case in this work, but are also often the starting point for other analysis methods, which calculate other parameters of biological interest.A parameter used frequently in dopaminergic imaging is the tracer-tissue binding potential (BP). TheBP is defined as the product of the Bmax, or receptor density, and the receptor affinity (inverse of the 31receptor-ligand dissociation constant KD) for a ligand in tissue.BP= BmaxK D =Bmax?affinitywhere KD may also be expressed as a ratio of binding rate constants kon and koff in a first order receptor-tracer binding relationship:dBdt =kon FR?k off BK D= k offk onwhere F is the concentration of unbound tracer, R is the concentration of free receptor sites, and B isthe concentration bound receptor-tracer complexes.Similar to the BP is the volume of distribution, which represents the volume of blood (plasma) that would contain the same amount of radiotracer as unit volume of tissue in a region being analyzed. The volume of distribution (DV) for tissue, VT, will depend on the concentrations of tracer in tissue, CT, and plasma, CP:V T=CTC PThe DV may also be related to rate constants for tracer uptake into tissue, with varying forms dependent on the tissue-tracer uptake model used (Innis, 2007).Both the BP and DV may be estimated through methods known as kinetic modelling. Kinetic models describe rates of transport of tracers between blood and a set of interconnected of tissue compartments that often represent different physiological states of a radiotracer (such as attached to a binding site, or free in tissue) that exchange the tracer at rates dependent on concentrations and rate constants for these exchanges. The time-activity curve, or time-dependent concentration in a region, of tracer, along with the blood concentration, are inputs to rate equations, from which rate constants or the BP or DV may be calculated.32The standardized update value (SUV) is another parameter that is frequently used for cancer imagingwith FDG in clinics. SUV is calculated as the ratio of the activity concentration in a region of interest over the injected activity per body mass of a patient. In this work, SUV is calculated for Mn-52 accumulation in rats after systemic injections, as discussed in the Multimodality Comparisons chapter.332 Tracer ProductionThis section contains a brief overview of PET radiotracer production, and a more-detailed discussion of Mn-52 and issues related to its use as a PET tracer. This includes irradiation of Cr foil to produce Mn-52, measurement of results of that irradiation, Mn-52 energy spectra measurements, and procedures for and results of separation of Mn-52 from the Cr foil in which it was produced.2.1 Background As discussed in the PET Introduction (see page 1), positron emission tomography (PET) uses positron-emitting radionuclides, often chemically bound into biologically interesting molecules, as the source of photons that are measured to acquire an image. In quantities small enough not to disturb the biological system being examined, such radio-labelled molecules may be referred to as PET radiotracers. Producing radiotracers involves two main steps: generation of the positron-emitting radionuclide, and radiochemistry to label the tracer molecules. This section briefly describes general PET radiotracers and their production.2.1.1 IrradiationRadionuclides for PET are most commonly produced in a cyclotron. Cyclotrons accelerate charged particles such as protons or negative hydrogen ions using oscillating electric fields between two half-circular electrodes, within a magnetic field that constrains the moving particles into a spiral path. Particles are injected near the centre of the cyclotron, and as they are accelerated by the electric field on each circuit around the cyclotron, they arc in increasing radius paths until they reach the outer edge of the system. Particles are then extracted and redirected along a beam line for subsequent use.For PET radionuclide production, negative hydrogen ions are accelerated in a cyclotron, stripped of electrons by a stripper foil to produce protons, which are then directed into a target material 34containing an isotope with a favourable reaction cross section at the energy of the protons for production of the desired radionuclide. The target may be a solid material, such as a metal foil, or may be a liquid that itself is the target, or a solution containing dissolved target atoms. PET radionuclide production reactions involving protons include O-18(p,n)F-18 (Bishop, 1996; Hess, 2001), N-14(p,?)C-11, and notably for this work Cr-52(p,n)Mn-52. In these reaction expressions, the bracketed symbols indicate the irradiation by a proton, p, and emitted neutron or alpha particle, n or?.Radionuclide production rates depend on the energy of the particles irradiating the target material, which for PET radionuclides are typically on the order of 10 MeV. The probability of a particular interaction is denoted by the interaction cross section, often expressed in barns (10?28 m2), a unit of area. Cross section, ?, may be converted into a linear attenuation coefficient, ?, for the irradiating particles in the target by multiplication with the density, ?, of target isotopes in the target volume, assuming the energy, E, of the particles is constant through the target material:?(E )=? (E) ?As with photon attenuation (see page 18), the total interaction probability for an impinging particle will depend on the thickness, t, of the target material, with an exponential decrease in flux with distance through the material, again under various simplifying assumptions such as constant particleenergy and no reduction in target particle density. In the case of a thin target with negligible attenuation of the beam passing through it, the instantaneous rate, R, of radionuclide production will be proportional to the thickness. The rate will also be proportional to the beam current, J, whichfor PET radionuclide production is typically on the order of 1 ?A:R=? (E) ? t JThe reaction cross section will vary with the energy of the irradiating beam, often with a complicated dependence with threshold energies and local maxima and minima (also much like photon attenuation energy dependence). For O-18(p,n)F-18, the reaction has a peak near 5 MeV with a cross section of 586 mb (Hess, 2001). For Cr-52(p,n)Mn-52, the reaction cross section increases with proton energy between 6 and 13 MeV, up to approximately 400 mb, then decreases with higher proton energies (NNDC EXFOR; Soppera, 2012).35For target objects that are thick enough to substantially reduce the energy of incident particles, the production rate will vary with distance through the target. As such, calculating the total production rate of an irradiation may require integration over the energies of the particles as they pass through the target.Additionally, depending on the relative decay and production rates of a radionuclide during irradiation, a saturation condition may arise where the rate of production is equal to the decay rate of the radionuclide, and no additional activity of the desired isotope will accumulate. For example, inF-18 production, the saturation activity increases with proton energy from 3 MeV to above 30 MeV, and is also proportional to the beam current (Hess, 2001).The TR13 cyclotron and its use for Mn-52 production in this work is discussed below.2.1.2 ChemistryAfter irradiation, most PET radiotracers will require one or more radiochemical procedures to prepare the tracer in a usable form. These may include separation of the radionuclide from unreacted target material, purification of a solution containing the radionuclide from contaminants, and chemical reactions to label a biologically interesting molecule with the radionuclide to produce the desired radiotracer.Radiochemistry for PET tracers generally requires rapid procedures due to the rapid loss of radionuclide activity due to radioactive decay (half lives of minutes or hours). Automated reaction systems are frequently employed to ensure rapid and reproducible reactions. The use of micro-fluidic systems, or "reaction on a chip", are a recent development as part of ongoing efforts to optimize this process. Automated systems also limit the radiation dose to radiochemists, and may beable to produce more consistent results by minimizing human variability in the process.After production, tracers are often subject to quality control testing, particularly when they will be injected in human or animal subjects.362.2 Mn-52The positron-emitting radionuclide Mn-52 is a primary subject of this work. Mn-52 has previously been proposed as a potential PET tracer for myocardial imaging (Chauncey, 1997), and has been used to study Mn absorption in humans (Davidsson, 1988). Prior to this work, its use as PET tracer had not been reported.This section discusses some of its properties that are relevant for use as a PET tracer.2.2.1 GeneralOne of the most notable properties of Mn-52 for its use as a PET tracer is its half-life of 5.6 days. Thisis an unusually long half life compared with standard PET tracers such as C-11 (20.4 min) or F-18 (109.77 min). The longer half-life of Mn-52 has some advantages; it allows imaging experiments to be conducted after a single injection over days or weeks, in which time tracers with half-lives measured in minutes would have decayed, allowing longer-term observation of biological processes.As well, the longer half-life allowed a more flexible schedule for radiochemistry to be conducted. However, the longer half also presents a challenge for radiation safety and storage. An in vivo injection of activity for PET imaging may remain present in the subject's body for days or weeks in amounts sufficient to require dedicated housing space and shielding to limit dose to workers.The energy of positrons emitted during the decay of Mn-52 is very similar to that of F-18. This range is short amongst positron-emitting radionuclides, potentially allowing Mn-52 PET to have resolution approaching limits of the scanning system, rather than being limited by the positron range.Alternate isotopes of Mn were considered. There are three positron-emitting isotopes of Mn with half-lives that are suitable for use as a PET tracer: Mn-52 (5.591 days), Mn-52m (21.1 min), and Mn-51 (46.2 min) (NNDC MIRD). Mn-52 was selected primarily due to its low (mean) positron energy and range in tissue (244.6 keV, 0.63 mm) (Le Loirec, 2007) which is comparable to F-18 (250 keV, 0.62 mm) (Disselhorst, 2010) and is significantly lower than Mn-52m (1179 keV, 5.288 mm) and Mn-51 (970.2 keV, 4.275 mm) (Le Loirec, 2007). The larger positron ranges of Mn-52m and Mn-51 would 37blur smaller image details, particularly in a PET scanner with resolution better than the positron range.A disadvantage of Mn-52 as a PET radionuclide is its branching ratio, or the number of positron emissions as a fraction of all Mn-52 decays. This ratio for Mn-52 is 29.6%, which means that for a nominal amount of activity, roughly 30% of the decays will emit a positron which can be detected by the PET system in order to produce images. Traditional PET tracers like C-11 and F-18 have branchingratios at or near 100%. The lower ratio of Mn-52 results in the need to use more Mn-52 activity that would otherwise be necessary to acquire the same amounts of data. It also has the consequence of increasing scanner dead-time effects without increasing usable data generation as quickly as would radionuclides like C-11 or F-18..2.2.2 Cascade GammasAnother disadvantageous feature of Mn-52 decay is the large number of cascade gammas that are emitted during a decay, as shown in figure 2.1. Due to the positron emission branching ratio of 29.6%, most Mn-52 decays release only a series of gamma rays with no positron. When a positron is emitted, there are also two or three gamma rays released with high-probability, which may interact with the scanner and generate a cascade background in acquired data (see pages 23 and 62).38Figure 2.1: Decay scheme of Mn-52 (NNDC MIRD).The large number of relatively high-energy cascade gammas released during Mn-52 decay also has the disadvantage of increasing the radioactive dose produced by exposure to Mn-52. Relative to other PET tracers which release only positrons, more high energy gammas are released by Mn-52 forthe same number of potentially useful (for imaging) 511 keV gammas. As well, the high energy gammas from Mn-52 decay are more penetrating of shielding materials than are 511 keV gammas. This may lead to increased radiation exposure to experimental workers, and may require additional layers of shielding material be used while storing wastes (see page 43).392.3 Mn-52 ProductionThis section describes the irradiation procedures for Mn-52 production in this work. The irradiation itself is discussed, as well as potential improvements to it that could be made in future work. The method and results of radioactive yield measurements are also given. Radioactive waste storage issues relating to of Mn-52 and Mn-54 are also discussed.2.3.1 IrradiationIrradiations were performed on TRIUMF?s TR13, a 13 MeV self-shielded negative hydrogen ion cyclotron, which produced a beam of approximately 12.5 MeV protons after passing through a target entrance foil. This beam irradiated natural isotopic composition chromium foil to produce Mn-52 through the Cr-52(p,n)Mn-52 reaction.The commercially purchased target foil was composed of natural isotopic abundance chromium (Cr-50=4.4%, Cr-52=83.8%, Cr-53=9.5%, Cr-54=2.4%; Goodfellow corporation) of 99.99+% chemical purity (ppm: Ag < 1, Al < 1, Ca 5, Cu 3, Fe 1, Mg < 1, Mn < 1, Si < 1), with 0.5 mm nominal thickness. The manufacturer's documentation noted that the actual thickness of the foil could vary over a single piece by up to 0.2 mm. The mass of most irradiated pieces of foil was not measured, and likelyvaried by up to 50% due to variations in the thickness, shape, and area of the pieces. Pieces were approximately 12 mm in diameter, irregularly shaped, with mass in one case of roughly 0.7 g. A pieceof Cr foil is shown in figure 2.2.40Figure 2.2: Irradiated Cr foil piece in beaker. Courtesy Paul Schaffer (TRIUMF).For irradiation, foil pieces were broken off from the single purchased foil by TRIUMF cyclotron operator staff. Chromium foil is quite brittle, and consequently great care was required to shape the foil and remove excess material so that it would fit in the foil holder apparatus. The foil could not be cut into the required shape for the usual method of clamping within a foil holder - elastomeric o-rings - which provide a seal for the helium cooling system. Instead, a 0.33 mm thick aluminum disc with a 15 mm diameter hole was used to hold the foil in place during irradiation. The target foil was cooled by helium jets on its front and back sides during irradiation, and the holder body was water-cooled. Helium is preferable for cooling the foil itself as the coolant is in the proton beam path, where water could attenuate and react with the beam, reducing proton flux on the foil and producing undesired radionuclides through reactions with oxygen in the water molecules.As 12.5 MeV protons stop in approximately 0.41 mm Cr (NIST PSTAR), the full beam power was deposited in the foil. Using estimates of the cross-sectional area of the beam on the foil (10 mm2), the beam power, the flow-rate of the helium (40 L/min), and a heat transfer coefficient between the foil and helium (2500 W/m2K), it was determined that a beam current of 2 ?A would raise the temperature of the foil by 154 K, which was deemed safe considering the melting point of Cr is 1857?C (Cornelia Hoehr, internal TRUMF safety proposal). Accordingly, irradiations took place at 2 ?Afor approximately 5 hours. The irradiated foil was allowed to decay for 10 hours prior to removal to allow any short-lived radionuclides to decay and to mitigate radioactive exposure to cyclotron operator staff during foil extraction.41Thinner Cr foil would have been preferable for the production of Mn-52 by proton irradiation because it would reduce the amount of Cr to be chemically separated, and because it would increasethe ratio of Mn-52 to Mn-54 produced. The range of 12.5 MeV in Cr foil (0.41 mm) is comparable to the thickness of the foil used in this work (0.5 mm), so the effective proton energy is somewhat less than 12.5 MeV due to losses while travelling through the foil. Threshold energies for Mn-52 and Mn-54 production are 5.60 MeV and 2.2 MeV (West, 1987), indicating that between these energies, onlyMn-54 is being produced. As well, the relative Mn-54 to Mn-52 activity has been reported to decrease with increasing proton energy for this irradiation (Dmitriev, 1968). Thinner foil was sought, but was not found commercially available without a polymer backing, which would have been problematic during proton irradiation. This supply issue is seemingly due to the brittleness of Cr foil brought about by oxide impurities, as high-purity Cr is reported as being an otherwise malleable metal (Stonor, 2002).Higher energy proton beams would be expected to improve the Mn-52 yield and Mn-52/Mn-54 ratioas the proton range increases with increasing proton energy. With longer proton range, less than thetotal beam power is stopped in the foil, potentially allowing a higher beam current to be used without sacrificing target cooling. Higher beam currents would also reduce irradiation times, which were up to 5 hours in this work. Published reports document the production of Mn-52 in 0.5 mm-thick Cr foil by irradiation with 17 MeV protons (range 0.7 mm) for 4 hr at a beam current of 10 ?A (Davidsson, 1988).2.3.2 Yield MeasurementsAfter proton bombardments, foils were dissolved in 10 to 20 ml of concentrated (12 N) HCl. Samples of foil solutions were taken by pipetting 10 or 20 ?l of the foil solution and mixing it into a glass vial containing 20 ml of tap water. Similar samples were also taken from solutions after subsequent radiochemistry steps, including the injectable purified Mn-52 solutions, and were taken while selecting and refining radiochemistry procedures, in order to assess the relative effectiveness of the separations for different methods.42The 20 ml Mn-52 sample vials were then placed inside an N-type coaxial high purity Ge gamma spectrometer (Ortec, Oak Ridge, Tennessee; shown in figure 2.3) fitted with a 0.5 mm Be window, and located at TRIUMF. This counter is routinely calibrated for energy and efficiency by TRIUMF staff using a 20 ml Eu-152 source, and standard calibration factors were used when analyzing data for thiswork. Gamma spectra were obtained for the Mn-52 samples by counting for 10 to 30 minutes.Figure 2.3: Gamma spectrometer shielded sample container and coolant supply.Spectra were analyzed by the counter software using peak detection with interference analysis to determine the activities of radionuclides present in the samples. This analysis uses a library of known radionuclide gamma emission patterns to identify the species in the counter and the amounts of each that are present. Interference analysis is used to account for cases where more than one radionuclide in the library emits gammas at energies that are close enough to be indistinguishable by the gamma spectrometer due to its limited energy resolution. For Mn-52, there are multiple high-probability gamma emissions in its decay scheme (744, 935, 1434 keV) in known ratios (90.0%, 94.5%, 100% of decays), making it a relatively robust radionuclide for this type of identification, compared with single-gamma emitters which could be difficult to distinguish from 43another source of a similar-energy emission. Notably, Mn-54 was also present in samples, but was less reliably identified in these fits, as its single gamma emission (835 keV) would occasionally be misidentified as arising from a different isotope.Knowing the volume of sample that was taken, the concentration in the sampled liquid could be determined. Similarly, when the sample was taken from a known total volume, the total activity in the sampled liquid could be determined. The results of these activity measurements indicate the yields of Mn-52 production after irradiation and after radiochemistry. For injections of Mn-52 in vivo,the concentrations injected were used with the known injected volume to estimate the total amountof Mn-52 that was injected in each procedure.Yield measurements on samples indicates that this irradiation procedure produces approximately 74.6 +/- 8.5 kBq/?Amin Mn-52 at end of bombardment. For a nominal 10 ?Ahr or 600 ?Amin irradiation, this corresponds to an average of 44.7 MBq of Mn-52 at end of bombardment. This is approximately 10% less than reported by Dmitriev (1968).2.3.3 Mn-54 Contamination / Waste StorageIrradiation of Cr foil with 12.5 MeV protons produces both Mn-52 and Mn-54. Natural Cr foil contains 2.4% Cr-54 which reacts with the protons, generating the radionuclide Mn-54 through the Cr-54(p,n)Mn-54 reaction.Yield measurements on samples (see page 41) confirmed the presence of Mn-54 (half-life 312 days) due to the presence of Cr-54 in natural Cr foil. Mn-54 activity was 0.56 +/- 0.16% of Mn-52 activity atthe end of irradiation. This Mn-54 contamination did not impact imaging, due to its relatively small amount, and because its single gamma emission at 834.8 keV is well-separated from the 511 keV positron annihilation energy.As noted above, increased proton energy or thinner Cr foil would reduce the ratio of activities of Mn-54 to Mn-52 from this irradiation. Similarly, enriched Cr-52 foil, or at least Cr depleted of Cr-54 relative to natural abundance, would be preferable for this irradiation, to reduce the amount of Mn-4454 generated. Acquiring such foil may be prohibitively expensive, however.Due to their half lives of 5.6 and 312 days, both Mn-52 and Mn-54 may present an inconvenience or problem for waste disposal and storage.The disposal limits for Mn-54 are such that the larger injections of Mn-52 in vivo in this work would be accompanied by higher Mn-54 concentration than may be disposed of as non-radioactive waste. For an animal sacrificed soon after injection, sufficient Mn-52 or Mn-54 may remain in the body to require decay prior to disposal. For example, a rat of body mass 490 g receiving an injection of 15 MBq of Mn-52 and 125 kBq of Mn-54, living for 6 days post injection and excreting approximately half of the injected Mn would retain about 60 kBq of Mn-54 after sacrifice. This carcass would need to decay to about 4.9 kBq Mn-54 before disposal as non-radioactive waste, which would take slightlymore than 3 years.For that reason, most carcasses of rats injected with Mn-52 / Mn-54 in this work were stored frozen for decay before disposal. For animals kept alive for weeks or months after injection, much of this would be passed in the feces into the into cage bedding, which may also present a waste disposal issue. Similarly, excess radiotracer solution, phantom objects after imaging was completed, and various consumables and waste materials from imaging sessions have the same issues. All these wastes required storage for decay before disposal.For wastes and carcasses requiring storage within the first few weeks after irradiation, the remainingpresence of substantial amounts of Mn-52 required additional precautions for storage (see page 38).Mn-52 in both carcasses and experimental wastes required shielding to reduce radiation fields. For frozen carcasses, an enclosure of lead bricks was constructed around the freezer to limit radiation fields in nearby work areas. Similarly, experimental wastes were stored under a lead enclosure. Live rats were housed in a specially approved room while radioactive (see page 82). Contaminated cage bedding was kept with the rat cages, behind a lead brick enclosure. Radiation fields were checked with a survey meter to verify shielding was adequate. Thickness of lead shielding varied, but was generally 5 to 10 cm.452.3.4 Specific ActivityAn important factor in many PET studies is that the radionuclide being used to produce signal is present only at tracer concentrations. That is, there is chemically and biologically negligible amounts of the chemical present, and the biological system being studied is not effectively altered by the introduction of the imaging agent. A key measure used to ensure tracer effects is the specific activityof the radiolabelled compound, which is the amount of radioactivity present divided by the mass or number of the tracer molecules (including both radiolabelled molecules and unlabelled carrier molecules in solution).To ensure tracer conditions are met, the total mass of injected chemical must be limited. With a low-specific activity radiotracer, the amount of activity that may be injected without breaking the tracer condition will be less. A limited injectable amount of activity may have implications for the statistical quality of PET image data that can be measured after the injection. For this work, the desire for a high specific activity tracer solution was motivated by desire to compare effects of injections of Mn at low and high total mass (see page 172).For Mn-52 in this work, specific activity is limited primarily by the Mn present as an impurity in the Cr foil prior to irradiation. As noted above, amounts of Mn in the Cr foil are at most 1 part per million, so for an approximately 1 g piece of Cr foil, at most 1 ?g or 18.2 nmol of Mn (atomic weight 54.9 ?g/?mol) is expected in the foil. Mass of the Mn-52 and Mn-54 radioactive atoms themselves isnegligible. For an average irradiation yield of 44.7 MBq, the initial specific activity would be 44.7 MBq/?g, 2454 MBq/?mol, or 2.5 MBq/nmol Mn. For comparison, Sossi (2007) reported specific activity up to 187 MBq/nmol (5045 nCi/pmol) C-11 of the dopaminergic radiotracer dihydrotetrabenazine (C-11-DTBZ).For systemic injections into animals (see page 82), up to 15 MBq was administered about 2 days after end of irradiation, or roughly 1/3 of the activity at end of bombardment. Precise measurements of Mn-52 radiochemical losses were not made, but assuming 1/3 was lost to radioactive decay (half life of Mn-52 is 5.6 days), the remaining 1/3 of the Mn in the initial solution 46was lost during radiochemistry, about 0.67 ?g of Mn from the foil would have been injected with theMn-52. In a volume of 2 ml, this is a concentration of 0.33 ?g/ml or 6.0 nM Mn.In comparison to the amounts of Mn given as an MRI contrast agent, these amounts of Mn from foil impurity are negligible. Systemic injections of Mn given as an MRI contrast agent, with or without Mn-52, involved amounts up to 20 mg in 0.5 kg animals. Direct brain injections of Mn were given with concentrations of 6.7 mM with Mn-52. As such, it is reasonable to describe Mn-52 solutions prepared for injection in this work without additional Mn as tracer solutions.2.4 Mn-52 Energy SpectraMn-52 decay energy spectra were acquired with the microPET scanner in order to investigate how that scanner detects the cascade gamma and positron annihilation photons from Mn-52 decay and to compare with similar spectra for the pure positron emitter F-18. These result were useful to inform method selection for future imaging experiments (see page 60), and to compare with the spectra measured with the dedicated multi-channel gamma spectrometer discussed above (see page42).2.4.1 AcquisitionA 30 ml polypropylene bottle (of approximately 3.5 cm diameter and 5 cm length) was filled with a solution of Mn-52 and Cr in HCl and water after a radiochemistry test separation. Initial activity was approximately 100 kBq/cc Mn-52. The bottle was placed inside the Focus 120 microPET scanner, within a 500 ml bottle of non-radioactive tap water. A similar phantom was independently prepared containing F-18.With a phantom in the scanner, data were acquired while adjusting the acquisition energy window. The Focus 120 energy window is set by specifying an upper and lower limit spectrometer channel. The channels range in number from 0 to 255, and are related to the measured photon energy:Energy (keV) = (channel #)*511/16047Spectra were acquired with 1, 3, or 5 channel windows, usually in increasing order of channel number starting with a window including channel 0. Data were acquired for about 10 minutes in each channel. The measured spectra were plotted as singles count rate (number of detected photons in energy window) against mean window energy. 2.4.2 ResultsThe Mn-52 and F-18 energy spectra acquired with the dedicated gamma spectrometer and with the microPET Focus 120 are shown in figure 2.4.Figure 2.4: Energy spectra of Mn-52 as measured with multichannel gamma spectrometer and with themicroPET Focus 120 system's adjustable energy window, and spectrum of F-18 measured with microPET.The Mn-52 gamma counter spectrum features distinct peaks for the 511, 744, 935, and 1434 keV cascade gammas, as would be expected for this radionuclide. In the microPET, the Mn-52 spectrum has broad peaks near 511 keV and 200 keV, with a local minimum between near 450 keV, and a largepeak increasing rapidly above a local minimum near 625 keV. The F-18 microPET spectrum has broadpeaks near 511 keV and 200 keV, with a local minimum near 400 keV and falling to near-zero above 650 keV.48The 511 keV peaks in the microPET spectra are associated with positron annihilation photons depositing nearly all their energy in the detector crystals, while the 200 keV peaks may be due to photons scattered in the detector crystal or in the field of view before depositing energy in the detectors. The F-18 spectrum has very few counts above 650 keV, which is consistent with there being no source of higher-energy gammas present. The Mn-52 spectrum has a large increase in events counted is seen above 600 keV, likely arising from a combination of the 744 keV gamma and 935 keV gammas scattered in the detector crystals or sample itself.2.5 Mn-52 RadiochemistryRadiochemistry was used in this work for isolation of Mn-52 from the Cr foil in which it was produced. This section describes the column chromatography procedure used for separation of Mn-52 from Cr after dissolution in concentrated HCl (shown in figure 2.5), as well as the work in development and testing of that procedure.Figure 2.5: Beaker of Cr foil dissolved in HCl on hot plate.Alternative Mn-Cr separation methods have been reported (Lahiri, 2006), however the method used in this work was selected because it was relatively simple and was adequate to prepare the radiotracers required for planned PET imaging experiments.492.5.1 Column ChromatographyThe radiochemical separations in this work were done by gravity column liquid-phase chromatography (illustrated in figure 2.6). This technique involves passing the solution containing the species to be separated through a bed of material that binds to some solutes more strongly thanothers. As additional solvent is added to the column, the solutes pass at different rates, spreading along the column length. Ideally, the individual solutes will elute in well-separated fractions and can be collected independently. In practice, there may be overlap between some solutes as they elute.Figure 2.6: Two chromatography columns eluting green Cr, with resin bed above withbulk of Cr passed, and Cr at top filtering through the column to drive elution.Details of the column preparation and its use that can affect the separation include the particular resin used, the volume of resin or mass of solutes, the manner in which additional solvent is added to the column, and how the solvent is forced through the column. For separations in this work, a 1 inch diameter class column was generally used, with a volume on the order of 200 ml. The column was driven by gravity, draining from the bottom under the solvent's own weight. Approximately 25 g of resin was used, which was dissolved in approximately 80 ml HCl, and further prepared by passing additional 80 ml HCl, and adding approximately 1.5 g acid-purified sand to the top of the column, before adding the solution to be separated. Further HCl was added in increasing volumes as the Cr-50Mn solution passed through the column bed, as the column would not drain without additional liquid above the column bed.2.5.2 Resin SelectionAnion exchange resins were used for separation of Mn-52 from bulk Cr dissolved in concentrated HCl. Two resins were evaluated by conducting test separations: BioRad AG 1x8 chloride form 200-400 mesh size resin (Bio-Rad Laboratories, California, USA), and Dowex 1x8 chloride form (Dow Chemical Company, Michigan, USA). Additionally, some AG 1x8 resin was converted from chloride form to acetate form, and this was used for an additional test separation. The test separations involved irradiated Cr foil containing Mn-52 which was dissolved and passed through a prepared bedof resin. The Cr and radioactive Mn were assessed in the column elutes to select the best performingseparation. Cr content in test elutions were assessed initially by visual inspection. Cr in solution produces a dark green opaque liquid, which fades to clear or yellow in later elution fractions, as indicated by the colours at the bottom of figure 2.7. A more precise technique was used for validation of Cr content later, but for initial tests, visual inspection was sufficient.Mn-52 content in test elutions was initially assessed by placing a vial containing a fraction (of volume2 to 6 ml) near a Geiger counter radioactive contamination meter, which would indicate the presence radioactive decays from the same with an audible clicking or hiss. Aural estimation of Mn-52 content was limited, so samples of the test elution fractions were also measured in a gamma spectrometer to more-precisely assess their radioactivity content, as illustrated in the graph portion of figure 2.7.51Figure 2.7: Mn-52 activity (curves) dependence on elution fraction number for Dowex and BioRad resinsfrom similarly prepared columns. Also shown are qualitative assessments of the colour of the elutions, which roughly indicate Cr content of the solution, with darker green colour indicating more Cr, and paler yellow colourindicating less Cr content. The BioRad resin has better separation between the peak of theMn-52 activity, and the range of fractions with visible indication of Cr content.For the purpose of isolating Mn-52 from bulk Cr, the two species would ideally elute from the column completely separately. However, only partial separation was obtained with both the BioRad and Dowex chloride form resins, with slightly better separation from the BioRad resin. In both cases, the Cr eluted first, as seen from the dark green colouration of the eluent, followed by Mn-52, in peaks with tails that decreased to below detectable radioactivity levels in later fractions. There was substantial overlap in the fractions that contained these species, however, with some of the Mn-52 eluting with the later fractions of the Cr peak, and non-negligible amounts of Cr remaining in some of the earlier Mn peak fractions.The additional acetate form test did not demonstrate useful separation of Mn and Cr, and was not pursued further. The motivation for trying the acetate form was the incomplete separation attained with the initial chloride form resins. After consultation with BioRad sales representatives, a 52document containing resin complexation tables was provided. These tables show plots of the approximate strength of binding between chemical species and resins for different concentrations ofsolvents. The complexation tables suggested that Cr-III would bind strongly to acetate form for the AG resin, and Mn would not bind at all. As noted, this did not result in the desired separation, however, and the attempt was abandoned.2.5.3 Cr Content TestingThe purpose of these separations was primarily to reduce the Cr content of fractions containing injectable Mn to acceptable levels.  The foil contained bulk Cr in amounts far greater than could be safely injected into animal subjects. No precise limit was set for Cr content in this work, although it was desired to keep it as low as possible to minimize potential complications. Notably, the Cr in the HCl foil solution was expected to be trivalent Cr3+, which is consistent with the dark green colour thatwas observed. Cr toxicity concerns are primarily with the hexavalent Cr6+ (as featured in the film Erin Brockovich (Universal Pictures, 2000)), which is carcinogenic. Conversely, Cr3+ occurs in many foods and intakes of up to 200 ?g/day are considered safe for humans (Porter, 1999).While radioactive Mn-52 could be tested with a gamma counter to accurately assess its concentration, non-radioactive Cr was more difficult to quantify. In particular, because Mn-52 is radioactive, samples that contained Mn-52 could not be sent for chemical lab testing. Instead, a mock separation was conducted with non-irradiated Cr foil scraps of similar volume to an irradiated foil. The dissolved foil was run through a similarly-prepared column, and samples were collected.The samples were sent for inductively coupled plasma mass spectrometry to assess the Cr and Mn content, with the Exova company (Surry, BC). The results of this assessment show steadily decreasing concentrations of Cr with additional eluent fractions (as shown in figure 2.8), and indicated an approximate volume of HCl after which the levels of Cr in the elute fell into the range of?g, which was deemed sufficiently low to be injectable. Future in vivo injections of radioactive Mn were prepared similarly, to minimize the amount of residual Cr injected.53Figure 2.8: Plots of Cr content and Mn-52 activity in similarly-prepared elution fractions. Some overlapof Cr and Mn content is seen, but Cr is largely absent by the peak of the Mn content curve.2.5.4 Separation ProcedureA Mn-Cr separation procedure typically involved the following steps:1) Prepare column:* Mass approximate 25 g of AG 1x8 chloride form anion exchange resin in a ~100 ml beaker.* Add approximately 80 ml of 12N HCl, stir into slurry.* Mount a 1 inch diameter glass chromatography column vertically in the fume hood, with spigot on bottom, open.* Pour resin slurry into vial, allow to drain until drips stop.* Pipette additional HCl into top of column to wash sides, and pour additional HCl to a total of 80 ml into vial, allow to drain.* Add approximately 1.5 g of acid-purified sand to top of column, rinse sides with a few ml of HCl.542) Prepare Mn-Cr solution:* Collect the irradiated foil from the TR13 cyclotron.* Place foil in a 100 ml beaker on hot plate.* Add 10 ml of 12N HCl.* Turn on hot plate to heat and dissolve foil, occasionally agitating beaker to swirl contents.* When contents nearly boiled dry, allow to cool, and add additional 10 ml HCl to redissolve.* Collect "stock" samples - typically 10 ?l - by micro-pipette, into 20 ml water vial.3) Run column:* Pipette 2 ml of Mn-Cr solution onto the top of the column, allow to absorb into bed so no liquid remains on top.* Repeatedly pipette additional 2 ml of 12N HCl to top of column to rinse sides.* Increase volume pipetted as visible green moves away from top of column.* After approx 30 ml HCl, green liquid begins to elute.* After approx 42 ml HCl, measurable radioactivity begins to elute.* After approx  54 ml HCl, begin collecting eluent for further processing.* Add additional HCl to top of column to wash out activity.4) Concentrate:* Combine eluents from multiple columns into single large beaker.* Place beaker on hot plate, and boil to dryness, which may take hours.* Cool, then redissolve in approx 3 ml PBS.5) Filter:* Collect Mn in PBS in syringe.* Attach 0.2 um syringe-tip filter.* Pass liquid through filter into sterile bubble-top vial.* Squirt small volume into fresh beaker.* Collect 10 ?l samples of small volume, into 20 ml water vial, for activity assessment in gamma spectrometer.552.5.5 Separation ResultsThe results of this separation are a pale-yellow to colourless, clear liquid containing approximately 30 MBq of Mn-52 in PBS. Activity concentrations up to 12.2 MBq/cc were produced after separation and filtration.These Mn-52 production and separation results were adequate for the purposes of this work, and allowed both intraperitoneal and intra-cerebro-ventricular injections in live rats. However, some complications occurred in these studies, as discussed in the Positron Emission Tomography Imaging chapter, which may have been partly related to the purity of the tracer solution. As noted above, a thinner Cr foil target and higher energy proton irradiation would improve yields and reduce the waste Cr to be separated and amount of Mn-54 contamination. Alternative radiochemical separationmethods should also be explored to improve the safety and consistency of the tracer solution. As noted above, alternative Mn-Cr separation methods have been reported. The separation method used in this work was time consuming, involved extended exposure to radiation, use of corrosive acid and hazardous chemicals, and was difficult to precisely reproduce. More rigorous quality controland production by more-experienced radiochemists using automated equipment and processes could likely improve the final product, its consistency, and take less time.563 Positron Emission Tomography ImagingPositron emission tomography (PET) experiments were an essential part of this work. One of the major goals was the development and testing of Mn-52 as a PET tracer, which required that PET imaging be conducted. This included initial proofs-of-concept, various image quality tests, and reconstruction validation on phantom objects. Experiments in live rats were also conducted to assess the biodistribution of Mn-52, to identify or verify potential applications for the tracer, and to provide data for comparison with magnetic resonance imaging (MRI) results.This section describes the UBC microPET Focus 120 PET system, and various phantom PET experiments that were conducted and their results. These included phantom studies conducted to assess Mn-52 PET image quality and compare with the established PET tracer F-18, and to develop or test correction methods used in reconstruction of Mn-52 PET data (see also page 23). PET experiments in live rats are also described in this section, including discussion of complications that arose and results that were observed.3.1 Focus 120 microPET ScannerPET imaging in this work was conducted on the Focus 120 dedicated small animal microPET tomograph, manufactured by Siemens Medical Solutions (Kim, 2007), which is shown in figure 3.1. The Focus 120 has 96 detector blocks containing 12x12 arrays of 1.5x1.5x10 mm3 lutetium oxyorthosilicate (LSO) crystals for a total of 13 824 crystals coupled to position-sensitive photomultiplier tubes. The scanner's field of view (FOV) is 10 cm radially and 7.6 cm axially, and the bore is 12 cm in diameter. The average energy resolution of the system is 18.3%. For acquiring transmission data to produce attenuation maps, a rotating point source holder is mounted inside thebore, which in this work held a Co-57 point source during attenuation scans. A rigid animal bed is mounted on a motorized stage that can translate into and out of the scanner bore, allowing subjects or objects to be positioned within the FOV after being placed on the bed outside the bore. Holes are drilled into the bed to allow mounting animal restraints or supports.57Figure 3.1: The microPET Focus 120 scanner, with 30 ml phantom bottle being scanned.The Focus 120 hardware is attached to a personal computer that runs the system control and acquisition software. Additional equipment in the vicinity of the scanner include an isoflurane anesthesia delivery system, which is used for initial knock-out and anesthesia maintenance during PET scans, and animal monitoring equipment (discussed further in the animal imaging; see page 82).3.2 Phantom ImagingPET images were acquired of phantom objects, which are designed to test the performance of a scanner, containing Mn-52 or F-18. These images were acquired in order to provide initial proof of concept for Mn-52 PET imaging, to understand how Mn-52 behaves as a PET tracer, to asses Mn-52 image quality in comparison with F-18, and to test corrections applied before data reconstruction. These phantoms and the metrics used to assess images of them are discussed below, and include a 30 ml bottle to assess cascade background; a specialized phantom to test several aspects of image quality; capillary tubes to test resolution; 5 ml vials to assess image calibration, time dependence, and contrast ratios; and a phantom used to acquire normalization data. These metrics were chosen primarily to test Mn-52 and compare it with F-18, and not for testing of the PET camera.58Also discussed below are details of the implementation of the cascade background correction method used in this work. To make Mn-52 PET useful for biological studies and to facilitate comparison between modalities, it was desired to achieve quantification of activity concentrations inreconstructed images accurate to within 10% after application of all corrections. This accuracy is comparable to scan-to-scan reproducibility with most PET tracers.3.2.1 Initial PhantomInitial experiments with phantoms containing Mn-52 solution were conducted to verify that PET datacould be acquired with the new radionuclide, and to determine how images would appear when using only standard correction methods that are designed for pure-positron emitting PET tracers. Of particular interest was how the cascade gamma rays produced by Mn-52 decay would affect imagingresults and how they would interact with changes to the energy window during data acquisition.To that end, one of the first objects scanned containing Mn-52 was a 30 ml polypropylene bottle, approximately 3.5 cm in diameter and 5 cm in length. This bottle was scanned containing aqueous Mn-52, initially at a concentration of approximately 100 kBq/cc, while placed within a 500 ml Nalgene bottle containing non-radioactive tap water, with the 30 ml bottle positioned near the centre of the scanner field of view, as seen in figure 3.2. In this arrangement, the larger volume of water provided attenuating material similar to the volume and mass of a rat.Figure 3.2: 500 ml Nalgene bottle containing non-radioactive tap water, and 30 ml bottlecontaining Mn-52. Bottle is on the microPET scanner bed in preparation for a scan.59For Mn-52 PET data acquired with the standard (for F-18 and C-11 tracers) energy window of 350-750 keV, a large cascade background was visible in images. This was particularly notable where the attenuation correction amplified the background signal in sinogram bins that passed through attenuating material. In cases where the bottle containing radioactivity was imaged inside the larger tap water bottle, the shape of water within that larger bottle became evident in attenuation-corrected images, as seen in figure 3.3.Figure 3.3: Single transaxial slice of 500 ml Nalgene water bottle and 30 ml bottle of Mn-52 solution.Data were acquired at 350-750 keV energy window, and reconstructed without cascade correction.The central orange circle is the 30 ml bottle, and the bluish solid shape around it is the outline ofnon-radioactive water that has been rendered visible by the reconstruction not accounting forcascade background. Colour bar scale ranges from 0 to 100% relative image intensity.An image acquired of similar geometry, with the 30 ml bottle containing F-18 (approximately 29.5 MBq at the start of a 1 hr acquisition), at the energy window of 350-750 keV shows essentially no background, even with attenuation correction applied, as seen in figure 3.4.60Figure 3.4: Single transaxial slice of 500 ml Nalgene water bottle and 30 ml bottle of F-18 solution.Data were acquired at 350-750 keV energy window. The central orange circle is the 30 mlbottle, and no discernible background appears around the bottle, unlike with Mn-52 atthe same energy window. Colour bar scale ranges from 0 to 100% relative image intensity.The background distribution seen in the Mn-52 image is problematic because it will contribute to voxel values throughout the image, and this contribution will not depend on the local activity concentration in the volume corresponding to the voxel. Instead, the contribution will be largely determined by the total activity in the scanner field of view, similar to the random coincidence background (see page 23), and the attenuation correction effect as seen above. Accordingly, it is necessary to remove or correct for this contribution to the image to produce voxels values that can be treated as proportional to the activity in the object being imaged.As seen in the energy spectrum data for Mn-52 and F-18 (see page 46), Mn-52 produces substantially more measured interactions in the Focus 120 system than does F-18 at energies outside the peak around the 511 keV positron annihilation energy. With a standard PET 350-750 keV energy window, these extra events may be included in coincidences detected between multiple crystals, leading to the background seen in the Mn-52 image.In order to study this effect, additional images of this phantom configuration were acquired with restricted energy windows, covering a smaller range of energies than 350-750 keV. It was hoped thatthe reduced range of accepted energies would reduce the fraction of coincidences recorded that involved photons arising from cascades. This was also expected to be useful because at the time of these experiments, a cascade correction method (as discussed below) had not yet been implemented, and it was not known if such a correction would be necessary for accurate 61reconstruction of Mn-52 data, or if the effect would be undetectable after only restricting the energywindow.The energy windows used included 400-650, 400-600, and 450-600 keV. The separate results are notnotable, and can be summarized by a trend that narrower energy windows resulted in less apparent background in qualitative inspection of the images. As discussed in the Tracer Production Mn-52 Energy Spectra section (see page 46), local minima appear in the Mn-52 energy spectrum near 450 and 600 keV, and a window bounded by these energies was considered a reasonable choice to effectively remove cascade photons while preserving detection of 511 keV annihilation photons. It was thus was decided to acquire subsequent phantom and live animal Mn-52 PET data with an energy window of 450-600 keV.Acquiring Mn-52 PET data with an energy window of 450-600 keV was qualitatively effective at reducing the cascade background in reconstructed images, as seen in figure 3.5. A rigorous evaluation of the optimal energy window for Mn-52 PET was not considered necessary for this work,as demonstration of the technique of Mn-52 PET could be conducted with the chosen window. Such an evaluation could be conducted by examining the noise-equivalent count rate (NECR) of the acquired data as the energy window is varied (Strother, 1990; Badawi, 1996), or by comparison of image quality metrics in reconstructed data similar to those discussed below in the Image Quality Phantom section (see page 70).62Figure 3.5: Single transaxial slice of 500 ml Nalgene bottle and 30 ml bottle of Mn-52 solution.Data were acquired at 450-600 keV and reconstructed without cascade correction.The restricted energy window has reduced the appearance of the background comparedwith the previous figure. Colour bar scale ranges from 0 to 100% relative image intensity.3.2.2 Cascade CorrectionRestricting the energy window as discussed above reduced but did not completely eliminate the appearance of cascade background in Mn-52 PET images reconstructed without cascade correction of the measured data. It was desired to produce the most background-bias-free and thus the most quantitatively accurate Mn-52 PET images as possible given the available measured data, in part to facilitate comparisons with other modalities. Accordingly, a cascade background subtraction algorithm was implemented and used as part of this work. Several methods have been published for this purpose including simulations (LaForest, 2009) and rescaled random coincidence distribution subtraction (Watson, 2008) (see page 23). Because it is simpler but effective, a rescaled-randoms method was used in this work.Conveniently, the histogramming step (see page 12) in the microPET Focus reconstruction software has an option to output a histogram of the randoms distribution that it estimates from the list mode measured data from a PET scan, along with the histogrammed emissions data. These outputs provided an easy starting point for cascade correction in this work, as they eliminated the need to create a custom randoms distribution estimation program, and provided the uncorrected emissions sinogram for input to a custom randoms and cascade correction program.Rescaling correction algorithms take a distribution that has the same shape as the contribution 63being corrected, and multiply that distribution by a scaling factor so that it matches the magnitude of the expected contribution to the measured data. For randoms-rescaling, the estimated random coincidence data in the radially peripheral histogram bins (those near the radial edge of the field of view) are used to determine a scale factor to match the measured emission coincidence data in those bins. That scale factor is applied to the entire estimated randoms distribution, and the scaled randoms data are then subtracted from all measured data histogram bins. This compensates for the contribution to the measured data from events involving cascade photons in all histogram bins. An illustration of this subtraction is shown in figure 3.6.Figure 3.6: Angle-averaged sinogram profiles for a single segment and axial offset. The uncorrected sinogramprofile, the rescaled randoms profile, and the cascade-corrected angle-averaged profile are shown.The use of peripheral sinogram bins to determine the randoms scaling factor can be justified by examining the sinogram images in figure 3.7 and sinogram profiles in figure 3.8.64Figure 3.7: Single scanner-axial-position slices from sinograms of PET data acquired on the Focus 120 microPET system. Plot axes are thehistogram bin radial offset (horizontal) and circumferential projection angle (vertical).  Scale units arbitrary. Top: The 30 ml phantom bottlefilled with F-18 with a 450-600 keV energy window. Centre:  A rat's head after IP injection of F-18, scanned with a 350-750 keV energywindow. Bottom: A rat's abdomen after IP injection of F-18, scanned with a 350-750 keV energy window.65Figure 3.8: Profiles through sinograms of 30 ml bottles containing F-18 (top, centre) and Mn-52 (bottom).Data were acquired with energy windows of 350-750 keV (top), and 450-600 keV (centre, bottom).Profiles were generated by averaging five sinogram planes (each similar to the left figure in theprevious set) and then extracting a single row (representing a single angular projection).For each profile, 20 central peak values were averaged, and used to normalize the profileto have a peak near 1, so that relative scale of backgrounds could be compared.66From the sinograms images, it can be seen that even with an object as large as a rat's head or abdomen in the Focus 120 scanner, there are 15 or more sinograms bins at the radial extremes which contain no peaks due to coincidences arising from unscattered positron annihilation photons. Instead, for objects similar to those scanned in this work, the contributions to coincidences counted in those peripheral histogram bins arise from other processes.The top and centre profiles in figure 3.8, both acquired with F-18, show the effect of restricting the acquisition energy window from the standard 350-750 keV to 450-600 keV (as is used for most Mn-52 imaging in this work). Outside of the central peak region where unscattered positron annihilation photons are counted in coincidence, the 350-750 keV profile has a radially-decreasing background contribution primarily due to coincidences involving a photon that has been scattered away from its initial path. The scatter contribution to the background is drastically reduced in magnitude and spatial extent in the 450-600 keV profile, likely due to a higher proportion of scattered 511 keV photons being rejected by the acquisition system due to the photons having lower energy after scattering. Assuming that a similar proportion of 511 keV photons are scattered and measured whenimaging with Mn-52 at the same energy window, it can be concluded that coincidences with scattered photons are a negligible contribution to the radially peripheral histogram bins, even when nearby bins have substantial counts due to unscattered coincidences.The bottom profile in figure 3.8, acquired with Mn-52 at 450-600 keV, demonstrates the relative magnitude of the cascade background contribution to sinogram data. Its magnitude is between 10% and 20% of the central peak region intensity for objects of this size, which is substantially higher than the fraction of coincidences involving a scattered 511 keV photon as seen in the F-18 images.Given these observations, it is reasonable to treat the contribution of coincidences involving scattered photons to the data measured in peripheral sinogram bins as low compared with the contribution of cascade coincidences. Accordingly, it should be a good approximation to use those peripheral bins to determine scale factors for fitting estimated cascade distributions to measured data for cascade background subtraction.67Estimated cascade background distributions, for use in correcting acquired Mn-52 PET data, were generated using program scripts written in MatLab (The Mathworks, Massachusetts, USA). The randoms, emission, and normalization sinograms were loaded using a script written by Raymond F. Muzic, Jr. (University Hospitals Case Medical Center, Ohio, USA). Normalization factors were applied to each histogram bin in the randoms and emission distributions to correct for system sensitivity variations. After normalization, the randoms distribution was smoothed by convolution with normalized 3 pixel wide uniform kernels in both axes of each sinogram plane. Smoothing was done to remove high spatial frequency variations from the randoms distribution that are most likely due to noise, and because the cascade coincidences distribution is expected to be relatively smooth across angles and radial offsets.To generate scaling factors to match the randoms distribution to the radially peripheral histogram bins of the emission data, the emission and smoothed randoms were first averaged over all circumferential projection angles to create one-dimensional profiles indexed by radial offset (as seenin figure 3.6). The outer 15 radial offset bin values on both sides (which as noted above primarily contain coincidences involving cascade photons) were then paired between the emission and randoms sinograms. Scale factors were calculated for each histogram angular segment and axial position from these ratios. The scale factors were then used to rescale the 2D randoms histogram foreach segment and axial position, which was then subtracted from the corresponding emission histogram data in order to correct for cascade events in the measured PET data. Scaling factors from this calculation varied between different Mn-52 PET scans, from averages near 12 to near 200, as is expected because the random coincidences are a variable fraction of the number of coincidences in a PET scan. Scaling factors for the first 95 sinogram bins, which represent planes perpendicular to the scanner axis, are shown in figure 3.9 for scans of a phantom object and of a rat brain. Within a single scan, the scaling factors typically exhibited variations of approximately 15% to 100% of their average value, with exceptionally large factors occasionally occurring on sinograms planes with indices near multiples of 24, corresponding to the transition between adjacent rings of crystal blocks in the Focus 120 system. These exceptionally large scaling factor peaks occurred in planes where the number of counts in the randoms estimate sinogram produced by the histogramming software were unusually low, compared with the majority of planes for a scan 68and amongst scans in general, thus the large scaling factors were not unreasonable. No unusual artifacts were observed in cascade-corrected sinogram data, suggesting there is no problem with thecascade distribution scaling for these planes specifically.Figure 3.9: Scale factors matching peripheral sinogram bins of estimated randoms distribution to measured coincidence data,plotted against sinogram plane number (axis labelled "Sinogram Bin") for first 95 sinogram planes, which are the planes thatare perpendicular to the scanner axis. Shown are data for a phantom (left) and rat brain (right) containing Mn-52.After subtraction of the cascade background from sinogram data, Mn-52 PET images in this work were reconstructed using the standard microPET filtered backprojection software, with scatter correction, unless otherwise noted. For the 30 ml phantom in the 500 ml Nalgene bottle discussed inthe previous section, an example image reconstructed with cascade correction is shown in figure 3.10.Figure 3.10: Single transaxial slice of 500 ml Nalgene bottle and 30 ml bottle of Mn-52 solution.Data were acquired at 450-600 keV and reconstructed with cascade correction.Appearance of background outside the central peak region is reduced by the correction.Colour bar scale ranges from 0 to 100% relative image intensity.69Images of that same phantom, filled with Mn-52 solution, with acquisition energy windows of 350-750 keV and 450-600 keV, and reconstructed with and without cascade correction as seen above were further analyzed by placing radial profiles through the images, which are shown in figure 3.11. These profiles illustrate the effect of acquisition settings and use of cascade correction, as discussed above, on the appearance of cascade background in the images.Figure 3.11: Radial profiles through 500 ml Nalgene bottle and 30 ml bottle containing Mn-52 solution.Profiles are shown at energy windows of 350-750 keV without cascade correction, and at450-600 keV without and with cascade correction applied before reconstruction. The combinationof the restricted energy window and cascade correction is very effectiveat removing cascade background in reconstructed images.For that phantom geometry, the uncorrected cascade background acquired with a 350-750 keV energy window produced images with background across the image. This included a prominent ring at 30% of the central peak activity at the periphery of the image, and a smoother background between 5% and 20% of the central peak in the space between the periphery and central peak. Withthe energy window restricted to 450-600 keV during acquisition, but still no cascade correction, the peripheral ring disappeared, but there was a background of between 0 and 5% of the central peak outside of the central peak. With cascade correction on the 450-600 keV data, the background level appears centred around 0 intensity, with localized peaks between +4% and -4% of the central peak intensity. Particularly because the background average was near 0, this was judged to be an effective70means for removing bias to voxel activity values from cascade coincidence background in Mn-52 PETdata.3.2.3 Image Quality PhantomIn order to better understand the performance of Mn-52 as PET imaging agent, a National Electronics Manufacturer's Association (NEMA) PET image quality phantom (NEMA, 2008) was produced by the UBC Physics machine shop. This phantom is design to assess several aspects of image quality, as discussed below, and has been used previously for comparisons between different PET radionuclides (Disselhorst, 2010).The image quality phantom was scanned in separate sessions containing Mn-52 (7.8 kBq/ml), and containing F-18 (9.7 kBq/ml) in the main compartment with volume of approximately 23 ml. Both images were reconstructed from approximately 4.7x108 true coincidences, on 0.433x0.433x0.796 mm3 voxels. The NEMA specification for analysis of images of this phantom (NEMA, 2008) was followed, including assessment of image noise, activity concentration recovery in small volumes, andspill-over of activity into regions of no activity. These metrics are discussed in detail below. Axial slices from these images are shown in figure 3.12.Figure 3.12: Axial slices from PET images of image quality phantom filled with Mn-52 (top) and F-18 (bottom).Uniform region (left), 1 to 5 mm diameter cylinders (centre), and water-filled (left side of right images)and air- filled (right side of right images) enclosures are shown. Both sets of slices are shown scaled relative to their own peak values. Colour bar scale ranges from 0 to 100% relative image intensity.71The image quality phantom contains a region (left column of figure 3.12) for testing variation of voxel values within a relatively large volume containing uniform concentration of a radiotracer. These variations may occur due to statistical noise, or may be due to systematic artifacts that arise during the acquisition and reconstruction procedures. Regions of interest (ROIs) of diameter 22.5 mm were placed over 12 planes in the uniform region of images of the image quality phantom to assess image noise (%STD) and to act as a baseline for the activity concentration recovery coefficients (see below) in the small cylinders. %STD is defined as the ratio of the standard deviations to the mean values of voxel values in these ROIs. Non-random structure in image value variations is assessed separately, as discussed in the normalization testing section below (see page 80).The phantom contains also activity-solution-filled rods of diameter 1, 2, 3, 4, and 5 mm surrounded by solid plastic which are useful for testing recovery of activity concentration in small volumes. Due to positron range and limited resolution of the imaging system, reconstructed image values from small sources tends to be lower than the values in the larger uniform region, with the same concentration of tracer in both volumes. The concentration recovery is assessed by the recovery coefficient (RC), which is a ratio of reconstructed concentrations in the small rods and larger uniformregion (and which, as a ratio of concentrations, is independent of the actual concentration in the phantom). Averaged images of 12 adjacent transaxial slices in the region with 5 parallel small cylinders were generated, in which the maximum pixel values were found for each cylinder, in order to determine activity recovery coefficients (RC) and their standard deviations (%STDRC). RC values arethe ratios of the maximum pixel values in the averaged image for each cylinder, divided by the mean of the pixel values in ROIs on the uniform region. %STDRC is calculated as:%STDRC=100?( STDlineprofileMeanlineprofile)2+( STDbackgroundMeanbackground )2where Meanlineprofile  and STDlineprofile are the mean and standard deviation of the single-slice voxel values in the 12 transaxial slices at the positions of the maximum pixel values for each cylinder in theaveraged slice, and STDbackground and Meanbackground are the standard deviation and mean of the voxel values in the ROIs on the uniform region.72The phantom also contains two smaller enclosures within the large uniform region, one of which is filled with non-radioactive water, and another which is filled with air, to test the appearance of background and spill-over from activity in the surroundings into non-active regions. Due to positron range and contributions such as random, scattered, and cascade coincidences, there is often a background in measured data that is not dependent on the local activity concentration in the object.ROIs of diameter 4 mm were placed over 9 planes in the air and non-radioactive water-filled volumes to determine activity spill over ratios (SOR). SORs are the ratios of the mean pixel values in these ROIs to the mean of pixel values in ROIs in the uniform region. For this image quality phantom,the enclosures have no activity, and would ideally have SOR of 0 if all backgrounds have been accurately corrected and no activity spills into the enclosures due to positron range.Results of these analyses are given below. Results from Disselhorst (2010) after scanning a similar phantom for 2 hours initially containing 3.7 MBq of F-18 with a Siemens Inveon small animal PET scanner using FBP reconstruction are also listed.The Uniform region ROIs had %STD of 3.2% for F-18 and 3.9% for Mn-52. Disselhorst (2010) reported%STD of approximately 6% for F-18.RC values for small cylinders are shown in table 3.1. The values reported by Disselhorst (2010) are similar for F-18.CylinderRadius (mm)Mn-52 F-18RC (%) STDRC (%) RC (%) STDRC (%)1 18 7 17 62 53 6 44 43 78 5 68 34 90 4 82 45 97 4 90 4Table 3.1: Mn-52 and F-18 activity recovery coefficients (RC) in small diametercylinders, relative to uniform region of image quality phantom.73The water and air-filled enclosures had spill-over ratio (SOR) of 5.0% and 3.7% for Mn-52 and 2.4% and 1.7% for F-18, relative to the uniform region activity. Disselhorst (2010) reported SOR values below 1% for F-18 for both cylinders. Higher SOR for Mn-52 is likely due to residual / uncorrected cascade contributions.Overall, the PET image qualities of F-18 and Mn-52 were similar as assessed using the NEMA image quality phantom. Little spill over was seen into void regions, and both had similar activity recovery insmall rods. This is consistent with the similar positron range of F-18 and Mn-52. The scale of background seen in images with this phantom with Mn-52, under 5%, is within the goal for quantitative accuracy for this work.3.2.4 Resolution PhantomTo assess and compare the resolution of F-18 and Mn-52 PET images, images of capillary tubes filled with solutions of those radionuclides were imaged, which is a standard method for assessing PET system resolution. A capillary tube holder was constructed by Ivan Klyuzhin (PhD candidate at the University of British Columbia). The holder has multiple holes drilled into it with 6 or 12 mm spacings, which rigidly hold 1.1 mm inner-diameter plastic capillary tubes parallel to the scanner axis. Separately for Mn-52 and F-18, 4 tubes were filled with radiotracer by syringe and capped with putty. The tubes were placed in the holder, which was in turn placed inside a larger cylindrical phantom that was filled with non-radioactive tap water (so that the positron range near the tubes was similar to that of biological tissue). The holes in which tubes were placed were arranged in an inverted T shape, with three holes arranged horizontally with a spacing of 12 mm, and a fourth hole 6 mm above the central hole. The phantom was placed in the scanner so that the central hole was near the scanner axis: 2.8 mm radially for Mn-52, and 0.8 mm radially for F-18.The capillaries were imaged in the microPET scanner, with 13x106 coincidences for Mn-52 and 71x106 coincidences for F-18. The images were reconstructed onto 0.108 x 0.108 x 0.796 mm3 voxels,without attenuation, scatter, or cascade corrections, as these do not appreciably affect resolution. Slices from these images are shown in figure 3.13.74Figure 3.13: Images of Mn-52 (left) and F-18) capillary tubes as resolution phantoms in microPET. Profiles areshown placed through the central peaks. Colour bar scale ranges from 0 to 100% relative image intensity.Single slices of these images of plastic capillary tubes containing Mn-52 and F-18 were fit with Gaussian curves along the profiles shown in the figure above, using the profile fitting tool in the ASIPro microPET image analysis software to determine the full width at half maximum (FWHM) of the peaks, which is a standard metric of resolution. Multiple axial slices were not averaged prior to the fitting due to a slight angle between the capillaries and the scanner axis, which would have blurred the summed image over multiple slices, and because the single-voxel-wide profiles were very smooth, so that statistical noise did not appear to be a concern.The fits FWHM of 1.83 mm (Mn-52) and 1.95 mm (F-18) for the peaks. These results indicate that Mn-52 produces microPET images with similar spatial resolution to those of F-18. This is consistent with expectation given their similar positron ranges. These results are also similar to previously-reported Focus 120 images of Na-22 point sources at a radial offset of 3 mm, with radial FWHM of approximately 1.95 mm (Kim, 2007) , and are similar to Siemens Inveon PET scanner images of capillary tubes containing F-18, with FWHM of 1.81 mm (Disselhorst, 2010).3.2.5 Calibration PhantomTo reconstruct images with voxel values in units of concentration, a calibration factor is required. In this work, such factors were determined by acquiring an image of a 5 ml glass vial containing Mn-52 solution, with initial activity concentration of approximately 6.5 MBq/cc. This concentration was independently measured by reserving a 20 ?l sample of the activity solution and measuring the 75activity using a dedicated gamma spectrometer (see page 42). PET images were acquired with the standard Mn-52 energy window of 450-600 keV, so that the scaling factor could be applied to other Mn-52 images acquired in the same manner. The calibration image was reconstructed with cascade correction, and circular ROIs were placed on 26 planes of the image, covering 1894 image voxels, of which one plane is shown in figure 3.14.Figure 3.14: Single transaxial slice of Mn-52 calibration phantom with ROI drawn.Colour bar scale ranges from 0 to 100% relative image intensity.The microPET Focus 120 ASIPro software uses these ROIs and a user-provided concentration to determine the calibration factor, which is then incorporated into the normalization data used for the initial reconstruction. Using these normalization data for subsequent reconstructions produces calibrated images. Calibration factors in this work, producing microPET images in units of Bq/cc wereapproximately 1.7x108 for F-18 (with a 350-750 keV energy window) and 2.9x108 for Mn-52 (with a 450-600 keV energy window). The differences in these factors are likely partly related to the energy window difference, and the difference in branching ratio for positron emission (0.967 for F-18 and 0.296 for Mn-52).It was desired to test the accuracy of using a single calibration factor for reconstructing images of objects containing different concentrations of Mn-52. To investigate this, the same vial used for calibration of Mn-52 images was scanned 13 times over 40 days with the same acquisition settings. Based on the initial concentration and known 5.591 day half-life of Mn-52, the concentrations in the vial at the times of these scans could be calculated. The images were reconstructed with the same 76(calibrated) normalization data for all scans.Intensity in images was found to decrease exponentially over a 100-fold change in Mn-52 concentration, with a least-squares fit slope of ?0.119 / day (R2 = 0.99983) to the natural logarithm of concentration plotted against decay time, as shown in figure 3.15.Figure 3.15 Natural logarithm of reconstructed activity concentration on Mn-52 phantom scanned repeatedly.This corresponds to a half-life of 5.822 days, 4% longer (i.e. slower decay) than the expected half-life of 5.591 days for Mn-52. The reason for this discrepancy in apparent half-life is not known, but may include dead-time effects being not properly corrected by the scanner software during reconstruction. A potential consequence of this deviation is an incorrect calibration of Mn-52 PET images, particularly when the activity in the scanner is much smaller (i.e. orders of magnitude less) than what was used during a calibration scan. Acquiring multiple calibration images might limit the impact of this variation, although this was not attempted in this project and is left for future work.A variation in calibration of this size would lead to inaccurate estimation of activity concentrations in images, dependent on the difference between the activities in the calibration and subsequent image. Bias would increase with difference in activities, reaching 10% when the imaged activity is approximately a factor of 12 larger or smaller than the calibration activity.773.2.6 Contrast PhantomAs discussed in the previous section, repeated scanning of the same phantom as its activity concentration decreased due to decay was useful for checking for deviations of reconstructed activity concentration from expectation as the amount of activity in the scanner changed. Equally or more important is whether regions of different concentrations in a single reconstructed image are reproduced with the correct ratio. Accordingly, it was desired to verify that contrast between different concentrations of activity in the field of view in a single scan was reproduced in the reconstructed image. This result is important if relative distributions of activity in objects are to be assessed, and particularly if they are to be compared with other imaging modalities.To that end, the calibration phantom discussed above was also scanned simultaneously with anothersimilar phantom in the field of view. This second phantom had a diluted solution of Mn-52, with activity concentration 26.56% of the calibration phantom concentration, again assessed by sampling and gamma counting. ROIs were drawn on the vials in images, as shown in figure 3.16, reconstructedwith and without cascade correction. The ratios between the intensities of the vials was 27.92% without and 26.53% with cascade correction, which are +1.36% and -0.03%, respectively from gamma counting result.Figure 3.16: Single transaxial slice of PET image of two glass vials containing different concentrationsof Mn-52. ROIs are drawn on each vial. Colour bar scale ranges from 0 to 100% relative image intensity.These results suggest that the contrast of in Mn-52 concentrations in objects in PET images is 78accurately reproduced, particularly when cascade background correction is used. The 1.36% higher ratio of concentrations without cascade correction is consistent with its expected effect; a small positive addition (e.g. the cascade background) to two positive numbers (e.g. the concentrations in the vials) will bring their ratio closer to 1.3.2.7 Normalization PhantomNormalization scans are routinely acquired on the UBC microPET Focus 120 system for use in reconstructing images acquired with energy windows of 350-750 keV with the tracers F-18 and C-11.This data is processed with a component or model-based algorithm using symmetries of the system geometry to generate normalization sinograms.For the Mn-52 scans in this work, an energy window of 450-600 keV was used, so specialized normalization data were needed. Additionally, F-18 and C-11 are also both clean positron emitters with no cascade background, so are generally able to use a normalization data set acquired with either tracer interchangeably. For Mn-52, the presence of a cascade background in the data could potentially affect the results of normalization in a manner that could make normalizations generatedwith F-18 or C-11 not applicable to Mn-52 imaging, or could adversely affect the quality of a normalization generated with Mn-52 if assumptions in the processing of the data are violated when a background is present. It was thus desired to investigate how best to acquire and use normalization data when measuring Mn-52 PET data on the Focus 120. This involved acquiring data for a Mn-52-based normalization, and comparing with F-18-based normalization data when reconstructing other Mn-52 PET data. Normalization data were acquired for the microPET using the standard geometry: a 700 ml plastic Nalgene bottle with 3.4 cm radius. Bottles of this size were filled with approximately 10 MBq F-18, similar to standard microPET normalization scans, and separately with approximately 8 MBq of Mn-52. For both tracers, normalization data were acquired at 450-600 keV, to match the emission data acquired for other Mn-52 scans. These normalizations were used both used to reconstruct a microPET image of a 5 ml vial containing Mn-52, shown in figure 3.17.79Figure 3.17: The same single coronal slice through PET images reconstructed from thesame acquired data of calibration phantom filled with Mn-52, reconstructed withnormalization data generated with F-18 (left) and Mn-52 (right). What appears to bea normalization artifact appears under the black + in the Mn-52-normalized image.Colour bar scale ranges from 0 to 100% relative image intensities.The resulting images were similar in quality, but with less prominent artifacts in image reconstructedwith F-18 based normalization, as illustrated by the profiles in figure 3.18.Figure 3.18: Profiles along the axial direction of the same single coronal slice through PET imagesreconstructed from the same acquired data of calibration phantom filled with Mn-52,reconstructed with normalization data generated with F-18 and Mn-52.As seen in the figures, the F-18 based normalization produced a more uniform reconstructed image of Mn-52 PET data than did the Mn-52 based normalization. Accordingly F-18 based normalization data were acquired and used for all other Mn-52 images in this work.The ability to use F-18 normalization data in this manner is helpful for Mn-52 imaging, as it would have been more difficult to repeatedly prepare and measure normalization data using Mn-52. F-18 is80routinely available, and more easily disposed of than is Mn-52, particularly when diluted into a relatively large volume as is used for normalizations.The reason for the better normalization result with F-18 based normalization is unclear. As noted above, it may be due to assumptions or details of the normalization processing program, used for Focus 120 reconstruction, which are not met by the distribution of measured events in Mn-52 PET data.3.2.8 Further Normalization TestingHaving decided to use F-18 based normalization data for subsequent Mn-52 PET image reconstruction, it remained to be tested how uniform and well normalized images of phantoms containing Mn-52 images were, particularly in comparison with images of phantoms containing F-18 (which are also reconstructed using F-18 derived normalization data). Accordingly, images of 30 ml bottles containing F-18 or Mn-52, the same configuration as used for initial Mn-52 PET and cascade tests, were acquired at an energy window of 450-600 keV. The images were reconstructed using the same F-18-based normalization data. Profiles through these images are shown in figure 3.19.81Figure 3.19: Axial (top) and radial (bottom) profiles on near-central planes of PET images of bottles of F-18 and Mn-52.Axial (9.5 mm thick) and radial (31 mm thick) profiles were placed through near-central planes of thereconstructed images of the 30 ml bottles containing F-18 or Mn-52. Standard deviations of values within the central plateaus are 2.3% for F-18 and 3.9% for Mn-52 in the axial direction, and 3.2% for F-18 and 2.9% for Mn-52 in the radial direction. These results indicate that F-18 based detector normalization data used in Mn-52 data reconstruction produces images sufficiently uniform to see variations of approximately 10% over a uniform background. Additionally, the in vivo images in this work typically have much larger signal to82background ratios than do these phantoms with their uniform concentration distribution. The concentrated peaks of in vivo images and their interpretation should not be impacted by normalization artifacts or image noise on the scale observed here.3.3 Animal Mn-52 ImagingAn important goal of this work was to demonstrate proof-of-concept for Mn-52 as a PET tracer, to investigate its behaviour and performance in vivo, to discover potential applications, and to provide data for comparison with manganese MR imaging. Accordingly, PET scans were acquired after injections of Mn-52 into live Sprague-Dawley rats. Appendix A contains a list of all animal imaging subjects, including those discussed in this chapter.3.3.1 GeneralBefore animal imaging experiments were conducted in this work, it was necessary to have the study plans approved by the UBC Animal Care Committee (ACC). The application includes a list of experimental team members, descriptions of the animal subjects to used, the objectives of the research, the procedures that will be performed including applicable standard operating procedure documents, justification for use of animals including potential alternatives, and animal monitoring plans and criteria for humane endpoints of animals showing signs of distress or illness. An initial pilot application was submitted and approved for non-radioactive Mn MRI imaging. Subsequently, a longer and more complicated protocol was submitted, revised in response to ACC provisos, and approved for Mn MRI, Mn-52 PET and mixed non-radioactive Mn with Mn-52 IV and IP injections and PET and MR imaging. Later, an amendment to this protocol was submitted and approved to permit ICV injections. Details of these experiments are discussed below. Complications that occurredwere reported to the ACC as part of protocol renewals, and clinical veterinarians were consulted to advise regarding animal welfare.A typical live rat PET scan at UBC begins by anesthetizing the animal in a plastic box with circulating 5% isoflurane in air. Once unconscious, the animals are generally placed prone in an adjustable plastic head-holder device mounted to the scanner animal bed, with a nose cone and retractable 83bars that are inserted in the ears, which are designed to hold the head still and in a consistent position with the brain centred in the scanner field of view. For some of the Mn-52 scans in this work, however, the head holder was not used, as its design could not easily accommodate positioning the rat for some abdominal scans. In either case, a nose cone was used to deliver isoflurane for anesthesia during scans. A pulse-oximeter infrared probe was also attached to the hind-foot, and a rectal thermometer inserted, as seen in figure 3.20. Temperature, heart rate, and blood oxygenation were monitored regularly during scans. Rats were administered subcutaneous saline to maintain hydration and atropine to maintain heart rate, and were exposed to a heat lamp to maintain body temperature while under anesthesia during PET scans.Figure 3.20: Rat on microPET scanner bed. Pulse oximeter and rectal thermometer are in place.Snout is within a makeshift anesthetic nose cone used for this scan instead of the plasticcone used when the head-holder is mounted.Three types of injections were used for animal experiments in this work: intravenous (IV), intraperitoneal (IP), and intra-cerebro-ventricular (ICV) targeted at the right lateral ventricle in the brain.  Both IV and IP injections are systemic, introducing the tracer to the entire body via the blood, either directly (IV), or indirectly (IP) after absorption through the tissues and organs within the abdominal cavity. The uptake of a radiotracer by IV and IP injections are different on timescales on the order of minutes, however after hours or days, there is little difference in small animals in the pattern of MnCl2 between these modalities (Kuo, 2005). This is important, as in this work comparisons are conducted between Mn-52 PET images, one or more days after administration by IP and IV. IV injections are however preferable for imaging uptake in the abdomen immediately after 84administration, because it eliminates any delay for tracer absorption, and does not produce a localized concentration of tracer at the site of injection as occurs with IP administration that could lead to confusion when interpreting images. IV injections are also more useful for short-term uptake experiments because they are easier to conduct while an animal is positioned within an imaging system; the tail is more easily accessible when the abdomen if the latter is centred in a scanner. IV injections are more complicated, however, as they require a restrained or anesthetized animal, and require a patent IV line, typically in a tail vein, to be established, which requires an experienced animal technicians and equipment. Additionally, an IV injection is more likely to have complications such as health effects or death of an animal, which require IV injections to be more slowly administered slowly and with smaller volumes that are usable with IP injections. As noted below, acute complications occurs with IV injections in this work, which lead to preference of IP injections when possible.ICV injections, in contrast with the systemic injection routes, places the injection at a specific location within the brain, notably within the blood brain barrier (BBB). This is important because the BBB limits uptake of some chemicals into the brain from the blood after systemic administration (seepage 172). ICV is more complicated process than the IV or IP, however. For this work, ICV injection involved surgery by animal technician Rick Kornelsen, who is experienced with giving this type of injection to rats. The Magnetic Resonance Imaging ICV Preparations section contains a discussion of issues related to this method (see page 189).After Mn-52 injections, live rats were housed in the UBC Hospital Animal Resources Unit (ARU), in a room approved for radiation storage. Animals had water and food available ad libitum, with a daily 12 hour light / 12 hour dark cycle. Animals were monitored daily after procedures, according to ACC requirements. Lead shielding was erected around the animals' cages to reduce radiation levels outside the room to acceptable levels. After rats were sacrificed, radioactive carcasses were stored for decay in a freezer in the UBC Hospital Brain Research Centre (BRC), surrounded by lead shielding (see page 44). Contaminated rat cage bedding was stored for decay in the UBC PET suite radiation lab, behind lead shielding. 853.3.2 First Mn-52 IP InjectionInitial Mn-52 in vivo images were acquired as proof-of-concept for Mn-52 as a PET tracer, and to begin assessing how much Mn-52 activity would be needed for useful imaging results to be acquiredafter systemic injection. At the time of this scan, the Mn-Cr separation chemistry was still under development (see page 48), and only a relatively small amount of radioactivity, 600 kBq of Mn-52, was available in injectable form. This activity was administered by intraperitoneal (IP) injection to a conscious male rat.That rat was scanned in the microPET Focus 120, 1 day after and again 2 days after the injection. Thistime delay was chosen to be similar to the timing of Mn MRI scans of the brain in rats, which typically are acquired 1 or more days after systemic administration. It was initially planned to scan the head of this animal, so the rat was mounted in the rat head holder. However, a coincidence detection rate of approximately 20 per second was reported by the PET scanner with the head centred in the FOV, which is orders of magnitude smaller than a typical scan, and was judged unlikelyto produce usable PET images. The scanner bed was then moved in as far as possible, to position as much of the rat's abdomen in the field of view, as the IP injection was into the abdomen and it was expected that more of the activity would remain in that region. Indeed, after repositioning, the count rate increased to approximately 70 per second, which while still very low, was better than the head position. A 1-hr emission scan was acquired at this position. On the next day, the rat was placed in a similar the scanner again, although without the head holder. The scanner reported approximately 20 counts per second centred on the abdomen at this time, and another 1 hour emission scan was acquired.Reconstructed images, shown in figure 3.21, unsurprisingly showed very low activity in the rat's body and low-quality of data compared with standard PET scans (and compared with subsequently acquired images discussed below). Activity was seen in multiple locations within the abdomen, however, indicating that some activity was accumulating away from the injection site. No anatomicalstructures were identified in these images due to their poor quality, but they did demonstrate that Mn-52 imaging in vivo was feasible, although larger amounts of injected activity would be required for IP injection.86Figure 3.21: First rat IP injection Mn-52 microPET images (red-white), overlaid on attenuation maps(grayscale background) to provide some anatomical context. Transaxial (left), coronal (middle)and sagittal (right) slices are shown. Images were acquired 21 (top) and 45 (bottom) hours afterIP injection. The PET images are not cascade corrected, and were not reconstructed withcalibrated normalization data, and were smoothed to improve visibility of features seen here.It was also noted that the radiation field from the body of the rat was comparable to or smaller than the field from the cage bedding. This suggests that a substantial fraction of the injected Mn-52 was excreted into the cage bedding in the feces, and not retained in the body at these time points after IP injection.3.3.3 First Mn-52 IV InjectionIt was desired to investigate the pattern in with Mn-52 was distributed in vivo immediately after an injection. To this end, a tail vein IV injection of 10 MBq of Mn-52 in 2 ml saline was given over approximately 5 seconds to a male rat under isoflurane anesthesia while that animal was in the microPET scanner. This relatively large volume injected was used to give as much Mn-52 as possible, due to limited ability concentrate the tracer solution. Approximately 5 minutes post-injection, the rat's heart and breathing stopped. Due to animal death, it was not possible to observe the time dependency of Mn-52 accumulation after injection with this animal. Consultation with clinical veterinarians suggested that this reaction was due to the speed of the injection, and later similar 87injections by tail vein IV of similar volume were done more slowly, and had no similar reactions. However, there had been approximately 5 minutes of biodistribution in which Mn-52 could spread through the rat's body, and the animal no longer required anesthesia to remain still in the scanner. As such, the opportunity was taken to acquire post mortem images of the rat which would not be possible with a live animal that requires anesthesia to remain still in the scanner and in which tracer distribution would be changing with time. 4 separate scans were acquired of this animal at different axial positions within the scanner, roughly described as the posterior, the abdomen, the chest, and the head. The images were offset axially by 80 image planes (63.68 mm), which left several axial slices of overlap between adjacent acquisitions. Each scan was 90 minutes in duration, and all were combined into a single composite image that covered the whole rat body. Such an image of a live rat is generally impossible to acquire, as a rat body is longer than the scanner field of view.The images were reconstructed separately, and combined into a single full-body Mn-52 PET image showing the distribution of Mn-52 after approximately 5 minutes of biodistribution after IV injection, shown in figure 3.22. Large accumulation is seen in the liver and kidney, and moderate amounts are seen in the throat, bladder and testicles. Lower levels are seen throughout the body, excepting the brain, which appears as a dark region surrounded by the head and skull activity, consistent with the BBB impeding short-term uptake in the brain.88Figure 3.22: Sagittal (top) and coronal (bottom) slices through PET image of rat body after Mn-52 IV injection.Image is a composite of 4 acquisitions of approximately 90 min each, after 5 min biodistributionpost-injection. This image was cascade corrected, but not calibrated.3.3.4 Second Mn-52 IV InjectionFor the reasons noted above, it was still desired to use an IV injection to observe the short term uptake of Mn-52 in live rats. Also as noted above, the IV injection procedures were adjusted after consultation with clinical vets, and no additional similar complications occurred as a result of this style of injection. A second IV injection Mn-52 experiment was thus conducted, in which a tail vein injection of 12 MBq of Mn-52 in 2 ml saline was given in 0.1 ml boluses every 30 seconds into a rat with body mass 326 g under isoflurane anesthesia.It was desired to observe the Mn-52 distribution during and immediately after injection in the abdomen, as previous results had suggested this is where much of the short-term accumulation would occur. It was also desired to observe the head and brain shortly after injection, to check for short-term accumulation in that region, to compare with MRI results after Mn administration. PET emission data were thus acquired of this animal starting at the time of the injection, alternating withthe scanner field of view covering the rat's head and abdomen. Scans were acquired of the abdomenat 0-30 min and 60-75 min, of the head at 30-60 min and 75-90 min. It was also desired to 89investigate the longer-term accumulation and removal of Mn-52 from the body of the rat. As such, this rat was scanned again for 30 or 60 min, centred on the head and abdomen, on post-injection days 1, 6, 14, and 21. The 1 day post-injection scan was acquired in part because that time point is roughly when Mn-related enhancement is strongest in MRI, and because it was the soonest after thefirst scan when it was considered safe to give a rat additional anesthesia. The subsequent scans wereacquired approximately once per week, to examine the longer-term biological removal of Mn-52 andloss of PET signal due to decay of Mn-52.This rat was also given a subsequent F-18 injection and scanned, as discussed below.Regions of interest (ROIs) were placed on the kidneys, liver, spinal column, and hip bones in Mn-52 images of this rat's abdomen acquired from 0 to 30 min after injection, reconstructed with 5 minute frames. Time-activity curves (TACs) were extracted from the voxel covered by these ROIs, and are shown in figure 3.23.Figure 3.23: Time activity curves (TACs) of abdominal regions of interest (ROIs)in Mn-52 image of rat from 0 to 30 min post-injection of Mn-52.Maximum intensity projection (MIP) images of this rat's abdomen at 0-30 min and 60-75 min are shown in figure 3.24. MIP is useful for these images to show uptake in the abdominal organs, as these structures are three-dimensional and cannot be captured in a single plane slice through an image, while an MIP condenses contributions into a single two dimensional image. Mn-52 uptake 90seen initially strongly in the liver, kidneys, and in a curve descending from near the lever that is identified as part of the intestinal tract. By 60-75 min, the activity in the intestines extends through more loops, suggesting passage of activity through the intestines after filtration through the liver and excretion in bile may be detectable with Mn-52 PET. Spinal column, hips, and leg bones are also faintly visible in these images at this time.Figure 3.24: Maximum intensity projections through abdomen of rat of PET images acquired afterIV injection of Mn-52. Images are shown from 0-30 min (left) and 60-75 min (right).Images of this rat's abdomen from scans acquired from 0-30 min and 6 days post-injection are shown in figures 3.25 and 3.26 after smoothing to improve visibility of features in the images. The spinal cord and hip bones appear distinctly. 91Figure 3.25: Coronal (left) and sagittal (right) slices of PET image of rat acquired for 30 min immediatelyafter IV injection of Mn-52. Isotropic Gaussian smoothing with FWHM of 1 mm was applied.Figure 3.26: Coronal (left) and sagittal (right) slices of PET image of rat 6 days after IV injection of Mn-52.Isotropic Gaussian smoothing with a FWHM of 1 mm was applied.The details of the timing of this bone uptake of Mn-52 are not known, however. The uptake may occur immediately after injection and stay where it initially accumulates, which may provide a measure that is sensitive to short-term bone metabolism, while also providing a longer-term means to track changes in bone structure with a fixed marker. Alternatively, the Mn-52 may be recirculated out of the bone continuously, with potential to continue accumulating in regions with more bone metabolism.Activity is still visible in these bones two weeks post-injection, as shown in figure 3.27. This accumulation suggests that Mn-52 imaging may be able to provide a means to investigate bone function or disease in an animal over days or weeks after a single injection.92Figure 3.27: Coronal (left) and sagittal (right) slices of PET images of rat 14 days after IVinjection of Mn-52. This image was reconstructed without cascade correction and was notcalibrated. Isotropic Gaussian smoothing with a FWHM of 1 mm was applied.Sagittal, coronal, and transaxial slices through this rat's head from a microPET acquisition 75-90 min, 1 day, and 6 days after IV injection of Mn-52 are shown in figures 3.28 through 3.30.Figure 3.28: Transaxial (left), coronal (middle), and sagittal (right) slices through PET image of rat acquiredfrom 75 to 90 min after IV injection of Mn-52. This image was reconstructed without cascade correctionand was not calibrated. Isotropic Gaussian smoothing with a FWHM of 1 mm was applied.93Figure 3.29: Transaxial (left), coronal (middle), and sagittal (right) slices through PET image of rat 1 dayafter IV injection of Mn-52. Isotropic Gaussian smoothing with a FWHM of 1 mm was applied.Figure 3.30: Transaxial (left), coronal (middle), and sagittal (right) slices through PET image of rat,6 days after IV injection of Mn-52. This image was reconstructed without cascade correctionand was not calibrated. Isotropic Gaussian smoothing with a FWHM of 1 mm was applied.Uptake is seen throughout the head, neck, and anterior thorax. Notably, a prominent void appears inthe image at the location of the brain, surrounded by accumulation in the skull. This is consistent with Mn in the blood not passing the blood-brain barrier (BBB). A bright localization of activity is seen in the pituitary gland, inferior to the void of the brain, which is expected because the pituitary is outside the BBB. Activity is seen inferior and superior to the spinal cord in the sagittal image, delineating its shape clearly. Activity is also seen in the throat, possibly the salivary glands or thyroid.Accumulation is seen in the lower jaw, likely at the site of tooth growth. Activity is also seen throughout the anterior of the head, in the snout and possibly olfactory organs. Regions of interest were placed over the brain and the pituitary in the 1 day post-injection image. These are discussed in the Multimodality Comparisons Mn Dose effects section (see page 222).943.3.5 IV F-18 Water InjectionDue to the appearance of structures that resembled bones in Mn-52 PET images of the second IV-injected rat, it was desired to compare those images with PET bone scans acquired using a traditional bone PET tracer, F-18. For this tracer, no radiochemical reactions are used to attach the radioactive F-18 to a biologically interesting molecule such as flurodeoxyglucose (FDG), a commonly-used PET tracer. Instead, the F-18 is simply dissolved in water and treated to remove any contaminants so that it is safe to inject in vivo, and accumulates prominently in bones.That rat (now mass 421 g) received an IV injection of F-18 water, 76 days after its Mn-52 IV injection.The F-18 was provided by TRIUMF for this experiment, where radiochemist Mike Adam performed the necessary chemical purification. Approximately 37 MBq F-18 in 0.5 ml water was injected over several seconds while the rat was under isoflurane anesthesia in the microPET Focus 120 scanner. Abdominal emission data were acquired for 1 h starting at the time of injection, as shown in figure 3.31, and head emission data were acquired for 1 h starting 75 minutes after the injection, as shown in figure 3.32,  to acquire similar short-term uptake images as were acquired for Mn-52 IV PET immediate after injection.95Figure 3.31: Coronal (left) and sagittal (right) slices through abdominal PETimages of rat immediately after IV injection of F-18.Figure 3.32: Transaxial (left), coronal (middle), and sagittal (right) slices through headPET images of rat 75 min after IV injection of F-18 water.As anticipated, F-18 accumulation in bones immediately after injection appears quite similar to the Mn-52 distribution seen immediately, days, and weeks after its injection in the same animal. As F-18 water is a routinely used bone scan tracer, this corroborates the interpretation that Mn-52 is providing a longer-term marker for bone imaging. However, the biological process by which Mn-52 accumulates in bone is not clear from these images. In particular, the utility of Mn-52 bone imaging may depend on whether the Mn is tightly bound in bone within hours or whether Mn may exchange between different sites in bone and the blood.The Mn-52 images in figures 3.29 (28.6 x 106 true coincidences) and 3.30 (8.4 x 106 true coincidences) are shown after smoothing with 1 mm FWHM Gaussian kernels, while the F-18 image in figure 3.32 (5.5 x 106 true coincidences) is unsmoothed. Even with smoothing, the Mn-52 images 96have more apparent background noise than the F-18 images in areas outside of large accumulations of activity. This is likely due to the contributions of the cascade background to noise, which is not removed by cascade background correction. As well, the Mn-52 images have a smaller apparent resolution than does the F-18 image, with smaller details of the activity distribution being qualitativediscernible. This resolution difference may be an artifact of the appearance of noise in the image, or may be a consequence of the different energy windows (450-600 keV for Mn-52, 350-750 keV for F-18) used during acquisition, which would be expected to affect the maximum angle, and thus reduction in energy, of 511 keV photons that are scattered in the animal and still fall in the accepted energy window by the scanner. A wider spread of angles would be expected to introduce an effective blur into the image, possibly explaining the reduced resolution of the F-18 image.3.3.6 ICV Mn-52 InjectionsPart of the motivation for developing Mn-52 as a PET tracer was to examine its behaviour in the brain, as neuronal-activity-related uptake of Mn is discussed in Mn-enhanced MRI publications. However, Mn-52 was not observed in substantial amounts within the brain after systemic injections as discussed above. Additionally, there was interest among team members regarding whether Mn transport along the spinal cord could be observed after injection within the brain.Two rats (340 g and 347 g) were anesthetized with isoflurane and their heads shaved. The rats were mounted in a surgical head holder with attached stereotaxic injection rig. The rats' scalp was openedby incision, and a hole drilled through the skull to access the brain. The rats were given 30 ?l right lateral ventricle ICV targeted injections containing approximately 365 kBq of Mn-52. One rat was left to recover for 1 day before scanning. The other rat was immediately taken to the microPET scanner for data acquisition. Anesthesia was not maintained during transport, but the duration was less thanfive minutes, and isoflurane anesthesia was resumed immediately after transport at the microPET scanner.It was desired to keep the duration of anesthesia of the rat that was scanned immediately after injection  short, as it was expected that additional scans would be conducted on subsequent days in order to observe the variation in Mn-52 distribution with time. As such, data were acquired with the 97rat's head centred in the scanner from 10 minutes to 40 minutes post-injection, and with the rat's chest centred from 60 to 80 minutes post-injection. Transmission scans of 10 minutes each were alsoacquired before each of these scans. The head scan was acquired to examine how the injection distributed within the brain from the site of injection, and the chest was chosen to show the spinal column, as shown in figure 3.33.Figure 3.33: Sagittal slice of PET image of rat acquired 60 to 80 min after ICV injection of Mn-52.Image is centred on chest, and covers the posterior end of the brain and spinal cord extending back fromit. Image is reconstructed without scatter or cascade corrections, and was smoothed to reduce noise.Image colour scale is shown at right as a rough guide. This scale, different from most other images shownin this section,  was chosen to better emphasize the curve of activity extending posterior from thebrain (white blob at top centre), which appears to follow the curve of the rat spinal cord.An attenuation scan is shown in grayscale underneath the PET intensity for anatomical context.After the acquisition, the rat was removed from the scanner and left to recover. The rat was checked intermittently for 30 minutes after scanning was complete, but was discovered to have stopped breathing at some point during this time. After the animal died, it was placed back in the scanner and additional brain image data were acquired for 4 hours, as shown in figure 3.34. Short-term distribution appears to be dominated by transport within the interconnected ventricles from the intraventricular site of injection98Figure 3.34: Transaxial (left), coronal (middle), and sagittal (right) slices of PET image of rathead, acquired within hours of ICV injection of Mn-52. Approximate injection site is markedby the white arrow, and the olfactory bulb by the yellow arrow.The second rat, which was allowed to recover for 24 hours after ICV injection before microPET scanning, was scanned for 2 hours, as shown in figure 3.35. The duration of this scan was as long as possible given the need to schedule several other scans on the same day. After removal from anesthesia, this rat similarly died during recovery.Figure 3.35: Transaxial (left), coronal (middle), and sagittal (right) slices of PET image of rathead, acquired approximately one day after ICV injection of Mn-52. Approximate injectionsite is marked by the white arrow, and the colliculus by the yellow arrow.99The reason for the reactions of these two rats is not known. Other rats receiving ICV or IP injections prepared on the same day from the same stock of activity, but slightly diluted with addition of non-radioactive MnCl2 did not have this reaction. Other animals receiving ICV or IP injections of only non-radioactive Mn had no similar reaction. Trial ICV injections of eluent from chromatography columns (prepared similarly to Mn-52 injections but without adding the dissolved Cr foil containing Mn-52, asdiscussed in the Magnetic Resonance Imaging ICV Preparations section, see page 189) did not produce similar reactions.3.4 Animal Mixed Mn-52 and Nonradioactive MnCl2 ImagingOne of the goals of this work was the comparison of results of imaging Mn distribution in vivo with both PET and MRI, and with post mortem autoradiography (AR) images. In order to facilitate these comparisons, rats were simultaneously injected with a mixtures of Mn-52 for PET and AR and non-radioactive MnCl2 to produce contrast and relaxation rate changes in MRI. These rats were then scanned with PET and MRI as close together as possible, and with AR after sacrifice. Details of the PET imaging experiments are discussed in this section. Comparison of results between modalities are discussed in the Multimodality Comparisons chapter.3.4.1 Mixed ICV InjectionAs noted above, the distribution of Mn after direct brain injection was of interest due to the use of Mn as a contrast agent in MR brain imaging. Additionally, because the brain is a rigid structure within the skull, it is possible to coregister separately acquired images of a brain in order to directly compare the results. Such direct comparisons would be more difficult using abdominal scans. As such, the imaging of Mn-52 and MnCl2 with PET and MRI after a single injection containing both was an important test for comparison between modalities, and it was desired to acquire images with which to do such comparisons. Initial preparatory ICV injection experiments were conducted before the injections and imaging described here, which are discussed in the Magnetic Resonance Imaging ICV Preparations section (see page 189).100Two rats (407 g and 428 g) received surgical ICV right-lateral targeted injections of Mn-52. These injections contained 290 kBq of Mn-52, in solution with 6.5 mM MnCl2 in a volume of 30 ?l. The purpose of these injections was to simultaneously delivery Mn-52 and MRI Mn contrast agent into the brain of rats. After imaging, it was discovered that one of the injections missed the right lateral ventricle, and instead injected into nearby brain tissue. This misplaced injection resulted in a markedly different distribution of Mn seen in images. These rats received MRI scans before and after injection, and microPET scans after injection at similar times to the MRI scans. Immediately after the final MRI, rats were euthanized by carbon dioxide asphyxiation, and were transported to the microPET scanner and scanned for 6 or 8 hours. These images are shown in figures 3.36 and 3.37. After scanning, rats were frozen for storage, but several weeks later rats' brains were also extracted and imaged using autoradiography.Figure 3.36: Sagittal slices through PET images of rat brain 1 and 5 days after ICV injection ofMn-52 and additional non-radioactive MnCl2 targeted at the right lateral ventricle.Figure 3.37: Transaxial (left), coronal (centre), and sagittal (right) slices through PET images of rat brain 8 days aftermisplaced ICV injection of Mn-52 and additional non-radioactive MnCl2 targeted at the right lateral ventricle.101After receiving injections, these rats both loss body mass over the following days. The rat with the successful injection lost 16% of its body mass after 5 days, and the rat with the unsuccessful injection lost 6% over 7 days, after which both were euthanized for scanning purposes. As discussed in the Magnetic Resonance Imaging ICV Imaging section (see page 191), similar injections of only non-radioactive MnCl2 in rats that received multiple anesthesias for imaging sessions also displayed stable or reduced body weight over similar time periods. Both these rats also developed forepaw redness and blackness which clinical vets suggested may be gangrene. This was judged by clinical vets to not require immediate euthanization, and the rats were treated with topical antibiotics and steroids for several days before euthanization. This reaction was not observed in rats given only non-radioactive MnCl2 injections.The misplacement of one of the two ICV targeted injections discussed in this section was unplanned,but in practice was useful in that it provided a very distinct distribution of Mn from that in the successful injection. Such misplacement is also not surprising; the routinely used method for preparing rats for other brain PET scans noted above involves injecting a neurotoxin into the subjects' brains to create lesions that unilaterally mimic damage caused by Parkinson's disease. This technique is not precisely controllable in practice, and attempts by the same technician with the same equipment and technique produce experimentally important variation in lesion severity. Similarly, in a separate set of injections of non-radioactive MnCl2 conducted later by the same technician, 1 of 2 similarly-performed injections was similarly misplaced. In addition to 3 other rats giving similar injections prior to the injections discussed in this section (see page 191), 2 out of a total of 7 rats given injections containing MnCl2 were observed to have misplaced injections. MR images acquired after lateral-ventricle-targeted ICV injection are shown in the Magnetic Resonance Imaging chapter, and illustrate the injection route and anatomy in this region.3.4.2 Second Mn-52 IP InjectionAs discussed regarding the Mn-52 IP and IV injection images, Mn-52 accumulation is not clearly seenin the bulk of the brain of rats after systemic injection. This is consistent with publications indicating that Mn does not effectively cross the blood brain barrier after systemic injection (see page 172). 102However, this result was somewhat surprising because multiple published MRI results, and MRI results in this work, show pronounced signal enhancement in the bulk of the brain after systemic non-radioactive MnCl2 injection, which suggests that some Mn does cross the BBB within days.The reason for this discrepancy between PET and MRI results was not immediately clear. An initial hypothesis was that the large amount of Mn injected systemically for MRI purposes was affecting the time course of its uptake into the brain. For PET Mn-52 imaging, conversely, orders of magnitudesmaller amounts of Mn are injected; Mn in PET injections is present mainly due to impurities in the Cr foil prior to irradiation that are carried through the radiochemistry process (see page 45). A hypothetical mechanism leading to this difference was a saturation of the liver's ability to filter out Mn from the blood when given large doses; in this case, more Mn would remain in the blood and have a longer opportunity to enter into the brain.It was desired to investigate these discrepancies, including the effect of a large dose of non-radioactive Mn systemically administered simultaneously with Mn-52, in comparison to the uptake of the tracer-dose Mn-52 in the IV injection discussed above. To that end, an injection was prepared containing both Mn-52 and non-radioactive Mn in amounts that produce pronounced signal enhancement in the brain in MRI after systemic injection. A rat (490 g) was given an IP injection containing 75 mM MnCl2 and 15.1 MBq of Mn-52 in 2.05 ml saline.Prior to the injection noted above, the rat was anesthetized and given a baseline MRI scan. The injection itself was given while the rat was in the microPET Focus 120 scanner, so that biodistributionof Mn could be compared with the second IV MnCl2 injection discussed above. The rat received MRI scans before and after injection, and PET scans after injection at similar times to the MRI scans. Immediately after the final MRI, 6 days after injection, the rat was euthanized by carbon dioxide asphyxiation, and was transported to the microPET scanner and scanned for 8 hours. It was then frozen for storage, and after several weeks, its brain was extracted and used for autoradiography. The distributions from these modalities were compared, as is discussed in the Multimodality Comparisons section (see page 209).103Slices from images of this rat are shown in the figures 3.38 through 3.42, in most cases after smoothing (as specified in captions) to improve apparent contrast and visibility of structures in image. Additionally, regions of interest were placed over the brain and the pituitary in the 1 day post-injection image. The mean values within these ROIs are discussed in the Multimodality Comparisons Mn Dose Effects section (see page 222).Figure 3.38: Transaxial (left), coronal (middle) and sagittal (right) slices of imageof rat brain 1 day after IP injection of Mn-52 and 75 mM MnCl2 in approximately2 ml volume. Isotropic Gaussian smoothing with a FWHM of 2 mm was applied.Figure 3.39: Transaxial (left), coronal (middle) and sagittal (right) slices of imageof rat brain 6 days after IP injection of Mn-52 and 75 mM MnCl2 in approximately2 ml volume. Isotropic Gaussian smoothing with a FWHM of 1 mm was applied.104Figure 3.40: Transaxial (left), coronal (middle) and sagittal (right) slices of PET image of ratabdomen brain immediately after IP injection of Mn-52 and 75 mM MnCl2 in approximately2 ml volume. Isotropic Gaussian smoothing with a FWHM of 1 mm was applied.Figure 3.41: Transaxial (left), coronal (middle) and sagittal (right) slices of PET image of ratabdomen 1 day after IP injection of Mn-52 and 75 mM MnCl2 in approximately 2 ml volume.PET emission data, with a red-white colour scale chosen for better visibility in this fused image thanthe rainbow scale used in other images in this section, are overlaid over attenuation map image(grey) for some anatomical context. The PET emission data were smoothed with an isotropic Gaussiansmoothing with a FWHM of 1 mm. The transmission scan attenuation map data weresmoothed with an isotropic Gaussian with a FWHM of 3 mm.105Figure 3.42: Sagittal slices of PET images of rat abdomen 1 day (left) and 4 days (right) after IP injectionof Mn-52 and 75 mM MnCl2 in approximately 2 ml volume. The 1 day data was smoothed with aGaussian with FWHM 2 mm, and the 4 day data was smoothed with a Gaussian with FWHM 1 mm.Scale of these images is deliberately compressed to emphasize a vertical (in the image) columnof activity to the right (in the image) of the large out-of-scale region. This appears tobe the spinal column, which is not visible at the wider scale seen in the previous figure.3.5 DiscussionAs discussed above, Mn-52 PET with phantom objects is broadly similar in image quality to F-18 (as assessed by the metrics discussed in the Phantom Imaging section, see page 57) which is not surprising given their similar positron energies. An exception to this is the appearance of cascade background in images, which necessitates a more restricted energy window and additional processing, but can be effectively removed before reconstruction. As noted, the optimal energy window for Mn-52 PET may not be the 450-600 keV window used in this work, and additional characterization of Mn-52 would be beneficial.In rats, Mn-52 PET images in the head have not been previously published, so cannot be directly compared with prior work. Mn-enhanced MRI results do show some similarities, particularly in the accumulation in the colliculus and pituitary, and some dissimilarities, particularly in the lack of uptake throughout the brain after systemic administration. These comparisons are discussed in moredetail in the Multimodality Comparisons chapter.As discussed above, Mn-52 PET appears to have potential applications independent of its use in MRI,particularly outside of the brain. Bone imaging is of particular interest, as the standard PET method 106for bone imaging uses F-18 as a radionuclide, which has a much shorter half-life and thus a more limited ability to study longer-term bone biology. Fore (1952) suggests Mn accumulation is in corticalbone, but not the marrow. Both F-18 and Tc-99m methylene diphosphonate (MDP; Peller, 1993) are cortical bone tracer used for fracture detection, cancer spread, and bone infections. To asses the utility of Mn-52 for longer-term bone imaging, it would be useful to compare it with Tc-99m MDP, and particularly to examine how distribution changes during fracture healing.Various complications arose during Mn-52 PET studies with live rats. Before conducting future Mn-52 PET in animals, improved irradiation and radiochemical methods should be investigated, as is alsodiscussed in the Tracer Production chapter.It would also be useful to investigate methods for blood brain barrier disruption in order to facilitate future Mn-52 PET studies in the brain after systemic injection. As discussed in the Manganese-Enhanced Magnetic Resonance Imaging Relaxivity Calibration section (see page 174), such methods have been employed for MEMRI investigations, and could also be useful for PET studies.Another particularly important future step for Mn-52 PET would be to attempt to measure activation-induced uptake. The motivation for Mn-52 PET was in part to take advantage of PET's ability to quantitatively measure concentration, which could be a very useful tool to more precisely measure patterns of neuronal activation in the brain.1074 AutoradiographyAutoradiography (AR) is a radiological imaging technique that predates positron emission tomography (PET). Autoradiography is considered to be a very reliable imaging method, and is used as a gold standard by which to validate other modalities such as PET and magnetic resonance imaging (MRI). Comparison with and validation of PET and MRI results was the primary motivation for acquiring the AR images discussed below.In the following text, background about AR and how it was done in this work is discussed, including the equipment and various aspects of preparation and acquisition of images, processing, and corrections. Lastly, result images are shown.As with PET, it was desired to achieve AR image values proportional to activity distribution in tissues within 10% accuracy, to facilitate comparisons between modalities.4.1 BackgroundThe term 'autoradiography' name refers to the origin of radiation from within an object, in contrast to external-source x-ray techniques that pass a radiation through an object to produce images. In autoradiography, like PET, a radioactive isotope is introduced into a sample to provide radiation for imaging. Unlike PET, autoradiography cannot be used to image activity in vivo; an autoradiography sample must be prepared as a thin slice and exposed to the imaging instrument in direct contact or within a few mm of its surface. Because the sample is thin and in direct contact with the screen, a high resolution image may be acquired without the need for a tomographic reconstruction.Traditional autoradiography produced images in radio-sensitive films such as those containing silver bromide and silver iodide emulsions, which are exposed to a flux of charged particles originating in the sample being imaged. The charged particles deposit energy in the film, causing the formation of progressively larger grains of silver ions which are later developed by a chemical process to produce an image in the form of variable optical density regions on the film (Lear, 1986). 108More recently, phosphor imaging systems have been adapted for autoradiography. These systems use reusable phosphor screens with higher sensitivity than film, which allows more rapid acquisition of images. This is particularly useful for autoradiography with PET radionuclides like F-18 or C-11, which have half lives of hours or minutes, which make exposures of more than one day impossible (Strome, 2005).Autoradiography experiments often expose tissue to a radioactive tracer solution ex vivo, similar to ahistological tissue preparation. Alternatively, a radiotracer may be introduced into a live animal and allowed to distribute, after which the animal is sacrificed and the tissue extracted while still containing measurable radioactivity. These methods may produce different results, as the biological mechanisms and environment in live tissue may be different from the uptake mechanisms ex vivo, including the presence of endogenous sources of the radiolabelled chemical.In this work, rat brain tissue was extracted and sliced after Mn-52 was injected into animals to act as a PET tracer, and that same Mn-52 was also used as the source of the autoradiographic signal. For those animals that also received MRI scans immediately before sacrifice, AR images could also be used as a standard with which to validate the MR results.Radioisotopes suitable for autoradiography are more varied than those suitable for PET, because the radionuclide decay path for AR need not include a positron emission. Autoradiographic images are most commonly produced by interactions of electrons arising in the tissue, which deposit their kinetic energy in the imaging apparatus. Autoradiographic images produced with PET tracer are formed similarly, with the positron depositing energy. The amount of energy of the charged particle may be on the order of 100 keV as with PET tracers, or may be less than 100 eV. In a suitable imagingsystem, lower-energy charged particles can produce autoradiographs with spatial resolutions less than 1 ?m (Lear, 1986). With positrons, the resolution will be substantially worse, due to the range of the charge particles in matter, much like positron range can limit the resolution of PET images. Strome (2005) reported autoradiographic spatial resolution of 470 ?m full-width at half-maximum ofthe image peak from a point source with F-18, which is likely similar to the resolution for Mn-52 due to their similar positron ranges.109A Cyclone storage phosphor system (Perkin-Elmer, Massachusetts, USA) was used in this work for reading phosphor plates after exposure. The Cyclone phosphor plates contain photosensitive crystalsof BaFBr:Eu2+. When exposed to radiation, the Eu2+ is reduced to Eu3+, producing conduction-band electrons which are trapped in Br vacancies. (Hillen, 1987; Perkin-Elmer Documentation). After exposure of the plates, the Cyclone system uses a solid-state red laser (633 nm) to cause the trappedelectrons to relax and emit 390 nm photons. The intensity of the released photons is proportional to the amount of radiation absorbed (with a linear dynamic range of five orders of magnitude) (Perkin-Elmer Documentation). This light is captured by optics and measured using a photomultiplier tube toproduce a 2D image of the phosphor plates that represents the distribution of radiation in the imaged tissue slices. The images have 42.3 ?m pixels (600 dpi), which can provide detail finer than any other modality in this work, although as noted above, the practical resolution of the images is often substantially larger than this distance due to positron range, and the resulting spillover between adjacent regions, particularly from areas of high to low activity in images (Strome, 2005).4.2 CalibrationCalibration of autoradiographs relates the intensity of the images to the concentration of radioisotope in the tissue being imaged. This is useful for experiments in which the absolute radiotracer distribution in tissue is important.Calibration is generally done using a series of standard blots placed on a slide during the sample preparation and imaged along with the tissue samples (Strome, 2005). For this process, drops of known volume containing a measured concentration of radioactivity in solution are placed on slides. The total activity in these blots is known, and can be used to construct a calibration curve that relates the intensity of the autoradiographic image to concentration. The intensities in image pixels representing tissue slices are converted to activity concentrations of the radioisotope labelling the samples.For this work, however, the animal subjects were injected roughly two weeks prior to the AR slice preparation, and no activity was available for creation of standard curves when the tissue slices wereprepared. As such, the results of autoradiography in this work could not be made quantitative in the 110sense of giving concentration of Mn-52 in the tissues.However, a typical autoradiograph using F-18 or C-11 has a very linear relationship between image intensity and calibrated concentration, so it is assumed for validation purposes that this relationship will hold for Mn-52 as well. The autoradiographic image intensities, measured in digital light units (DLU) by the Cyclone storage phosphor system, are thus taken to be proportional to Mn-52 concentration after background corrections as discussed below. The AR images thus give a relative measure of the distribution of activity in the imaged slices. Even without calibration, this provides a useful tool for validating the relative distributions of Mn measured with PET and MRI.4.3 Preparation and AcquisitionAfter rats were sacrificed, their bodies were frozen in household freezer surrounded by a wall of leadbricks for radiation shielding. Ideally, the brains of these rats would have been extracted prior to freezing, however time limitations due to the ongoing acquisition of MRI and PET images with these animals made this impractical until a later time. Instead, approximately 10 days after the Mn-52 injections, two of the rats (those which received ICV injections of Mn-52 without any additional non-radioactive Mn) were partially thawed to permit extraction of the brains. Brains were sliced into 20 ?m segments onto slides, which is the standard thickness used for other rat brain autoradiography preparations with the equipment that was used. The slices were oriented perpendicularly to the anterior-posterior axis of the brain, to roughly match the orientation of transverse slices in PET images.Slides were placed against radio-sensitive phosphor screens overnight to generate autoradiographic images. This duration of exposure is somewhat longer than the standard 4 hour exposure used for C-11. Due to its short half-life, C-11 would gain negligible benefit from longer exposures, while Mn-52 decay is slow enough that activity remains sufficient to continue increasing statistical quality of data weeks after administration. Additionally, the overnight exposure was chosen partly for practical reasons of research staff availability.After the initial one-day exposure, the first two rat brain slice images were read on the Cyclone 111storage phosphor system. A single slice image is shown in figure 4.1. This exposure duration was judged adequate upon visual inspection; small details of the activity distribution could be seen, the overall outline of the brain was visible, and statistical quality of the data did not appear to be a substantial limitation to interpretation.Figure 4.1: Raw autoradiograph slice from brain of a rat that receivedICV injection of Mn-52, cut perpendicular to the anterior-posterior axis.Colour scale ranges from 0 to 104 DLU (digital light units).Based on these good results, the remaining 3 rats' (which received IP or ICV injections of Mn-52 mixed with non-radioactive Mn) brains were sliced onto slides and imaged in a similar manner, although with slices oriented perpendicular to the superior-inferior axis in order to better match the slice orientation in already-acquired MR images of the same animals. The slides were again placed on phosphor screen and left to be exposed overnight, which was judged likely to be adequate for slices containing similar initial concentrations of activity. These images were acquired approximately 14 days after the initial injection, and read out using the same Cyclone storage phosphor system.The AR data of the rat that received a misplaced ICV injection of mixed Mn-52 and non-radioactive Mn was distinct from all others due to the larger dynamic range of Mn-52 concentrations it contained in tissue regions, and because the misplaced Mn-52 did not spread throughout the brain to the degree observed in the successful injections. A representative brain slice from this animal is shown in figure 4.2.112Figure 4.2: Raw autoradiographs of same slice of brain in a rat thatreceived misplaced ICV injection of Mn-52 and non-radioactive Mn.At left, scale is 0 to 5x102 digital light units (DLU), and at right, scale is 0 to 5x104  DLU.Instead, the activity was heavily concentrated near the site of injection. Within the peak regions, theimage intensities (measured in digital light units (DLU) by the system) in this animal's autoradiographwere about one order of magnitude higher than the highest peaks in other images, and two or moreorders above its own surrounding tissue. With a restricted display scale, structures of the brain such as the colliculus could be discerned as activity concentration peaks, outside of the spillover region near the main peak. This image were judged adequate for subsequent analysis, and had potential to be overexposed in the high activity region if left for a longer duration, so no additional exposures of this set of slices was acquired.The AR image of the rat that received the successful ICV injection of mixed Mn-52 and non-radioactive Mn had a different general appearance from the images discussed above. Rather than highly concentrated peaks (or a single peak) with relatively little between or around the peaks, this image showed a more even distribution. This is likely due to the longer time after successful ICV injection into the cerebro-spinal fluid (CSF) that the rat was alive and during which the Mn-52 could distribute throughout the brain.Because of its more-spread-out and less-peaky activity distribution, this image was more visibly impacted by statistical noise. In the initial 1 day exposure image, the noise was qualitatively moderate; there was sufficient contrast above the noise to discern structures in the image, but noisewas prominent enough to obscure some smaller structures. A second image of this set of slices with a longer exposure time was thus judged necessary.113Accordingly, the brain slices of the successful ICV-injected rat were exposed to a phosphor plate for asecond time, starting approximately 18 days after the initial injection, and left over a weekend, for a total exposure time of approximately 3 days (the precise timing of which was again determined in part by availability of research staff). The second image was qualitatively improved compared with the first for this animal, with better ability to see details due to an increased contrast to noise ratio. Comparisons of slice images with colour scales adjusted to appear similar are shown in figure 4.3.Figure 4.3: Raw autoradiographs of the same slice of a rat that received a successfulICV injection of Mn-52 and non-radioactive Mn-52. The left and right images wereexposed for 1 and 3 days, respectively. At left, the colour scale ranges from0 to 500 DLU, while at right, the scale ranges from 0 to 1500 DLU. The structuresvisible in the images have similar colours at these scales, but the longer exposure hasbetter ability to see details of the activity distribution due to reduced noise.Autoradiographic images of the rat that received an IP injection of Mn-52 and non-radioactive Mn were of drastically worse quality than the previously discussed images. This was likely due to the minimal uptake of Mn-52 into the brain after systemic injection in this animal. It was immediately judged necessary to acquire a second image with a longer acquisition. Slices from these images are shown in figure 4.4.114Figure 4.4: Raw autoradiographs of the same slice of brain in a rat that received an IP injection ofMn-52 and non-radioactive Mn. Both images have scale from 0 to 1x102 DLU.At left, the image was exposed for approximately 1 day starting 14 days post-injection.At right, the image was exposed for approximately 3 days starting 18 days post-injection.Even after acquiring a 3 day exposure, however, the statistical quality of the autoradiographs for the IP injected animal were qualitatively quite poor. Some structure is vaguely discernible, but it was notuntil the image was further processed (see below) that distinct structures in the activity distribution were readily visible.4.4 Background CorrectionThe images shown above were cropped to show the area immediately surrounding a single tissue slice of an autoradiographic image. In most of these cropped images, and particularly in the image ofthe IP injected animal, background noise can be seen in the areas outside the tissue slice itself. This background is a common characteristic of autoradiographs acquired with the Cyclone phosphor system, and also occurs with nearly-pure positron emitters such as C-11 and is observed even when the phosphor plates are not directly exposed to any source of charged particle radiation. This image intensity background may be an artifact of the process used to generate images from exposed autoradiography image plates, or may be related to uniform background radiation from the environment or photons produced by radioactive decay interacting with the imaging plate much further from the location of decay than do charged particles such as positrons.Regardless of the cause, it was necessary to correct for the background present in images in this work, as the background intensity reached values up to 25% of typical intensity values in images of brain tissue after ICV or IP injection. For comparison with other modalities' estimates of 115concentrations, ideally the intensity of images will be proportional only to the local activity concentration in the tissue, within 10%, and not depend on a non-locally-dependent background contribution to image value.4.4.1 Raw DataWhile the images shown above are cropped to display only single slices of the autoradiography data,the raw images actually contain impressions of approximately 46 separate brain slices arranged on two phosphor plates. Displaying a full two-plate image, and appropriately adjusting the colour scale as in figures 4.5 and 4.6, reveals structure in the background activity that is not readily apparent in the previous images.116Figure 4.5: Full raw autoradiograph of rat that received successful injection of Mn-52 andnon-radioactive Mn into the right lateral ventricle.Figure 4.6: Full raw autoradiograph of rat that received successful injection of Mn-52 andnon-radioactive Mn into the right lateral ventricle. This image is rescaled to emphasizethe appearance of the background distribution between the tissue slices thatappear as dark red regions, off high end of the colour scale.The non-uniform shape of the background between tissue slices seen in this figure presented a problem for analysis of the images. As noted above, calibration of autoradiography image intensities to concentrations of radioisotopes in tissue is typically done using a series of standard blots with known concentration of activity, but this technique could not be used in this work. That calibration process also accounts for the presence of background in the image, by placing a region of interest in the autoradiography image that does not contain the image of any local tissue activity, and including this point in the calibration curve. This method is generally adequate for autoradiographs using C-11,as the background in those images is much more uniform or flat than seen in the Mn-52 images shown here, and is reasonably approximated by a single constant background level or a single calibration curve for a whole plate. The cause of the different background shape with Mn-52 images 117from that seen with C-11 may be the presence of numerous higher-energy cascade gamma ray emissions from Mn-52, which do not occur with C-11, and may also be related to the longer exposure time used in this work for Mn-52 imaging (days, instead of hours for C-11).4.4.2 Constant SubtractionFor this work, it was necessary to develop background corrections that did not assume a single constant background level. That said, the first step for background correction for all autoradiographs in this work was to subtract a uniform background level from each plate of slices.A plate contains roughly 24 brain slices arranged in a 4 by 6 grid, as shown in the figures below. As far as possible from the tissue slice locations, tissue-slice-sized regions of interest were placed over the background plate images, examples of which are shown in figures 4.7 and 4.8. The pixel values inthese ROIs were averaged to determine the constant background contribution for each plate. The constant background was then subtracted from the intensities of pixels in the images, separately for each plate in each image.118Figure 4.7: Autoradiographic image of slices of brain of rat that received misplaced ICV injectionof Mn-52 and non-radioactive Mn. Blue rectangles indicate locations in which backgroundintensities were averaged for each plate to estimate the constant background contributions.The left and right plates had separate constant backgrounds calculated.Figure 4.8: Autoradiographic image of slices of brain of rat that received successful ICV injectionof Mn-52 and non-radioactive Mn, after subtracting constant background contribution.The intensities here are scaled to show remaining distribution of background between tissue slices.The residual background after constant subtraction can be seen in figures 4.7 and 4.8. The residual backgrounds in the images of rats that received ICV injections of Mn-52 (including both of the slice cut orientations discussed above) had similar patterns; near the locations of tissue slices, the residual backgrounds were highest, and fell with distance from the tissue slices. The residual background of the IP injected rat had a somewhat different structure than that of the ICV injected rats, as seen in figure 4.9. Instead of falling off with distance from slice areas, the background with the IP injected rat varied much more slowly across the autoradiographic image, without any simple dependence on the locations of the tissue slices.119Figure 4.9 Autoradiographic image of slices of brain of rat that received IP injection of Mn-52and non-radioactive Mn, after subtracting constant background contribution. The intensitiesare scaled to show remaining distribution of background between tissue slices. This image isdown-sampled more than the other rats' images due to fewer counts and resulting larger noisein the background, which makes it difficult to see the distribution without averaging pixels.4.4.3 Position-Dependent SubtractionBecause backgrounds remained visible in autoradiographs after constant subtraction as discussed above, and was of a size above 10% of the values in some tissue voxels, it was necessary to develop additional algorithms for background removal. Due to the different shapes of the background between the ICV and IP injected rats, different methods were used for autoradiographs of the animals that received these types of injections.4.4.3.1 IP InjectionFor the IP injected animal, in which the background was relatively slowly varying (compared with thesize of the tissue slices) and in which the background was not apparently strongly correlated with the intensity of the images within the tissue slices, a relatively simple second step correction was used. For each of these slices, a threshold value was chosen that qualitatively appeared to separate the activity within tissue slices and the surrounding background. For each single-slice, the surrounding pixels below the threshold were averaged, to determine a background level for that slice. Each slice's constant background was then subtracted from the corresponding slice. The result appeared to be a near-zero average background outside of, and presumably within, the tissue slices. 120An example slice is shown in figure 4.10.Figure 4.10: Single slice autoradiograph after background subtraction of rat that received IPinjection of Mn-52 and non-radioactive Mn. Image has been down-sampled to reducethe appearance of statistical noise and better reveal structure in the activity distribution.It would be reassuring to validate the results of this background subtraction process, but doing so is difficult at best due to lack of suitable metrics on which to base a judgement of the quality of those results. By design, the average background level after subtraction steps is zero, and the background contribution to pixel values within tissue slices cannot be directly determined. Comparisons of tissuepixel intensities between slices is complicated due to variations between slices due to actual concentration differences, as well as distortions and artifacts in slices images. Notably, however, the method described here is conceptually quite similar to the inclusion of a background component in the standard calibration method for C-11 autoradiography as discussed above, with the difference that the background is measured per-slice rather than just once for a whole plate. Additionally, as noted in the motivation for this technique, the background is qualitatively uniform for these slices, and this observation is consistent with post-correction images as well. Lastly, the purpose of this corrections is not to produce perfectly accurate images, but rather it is to remove as much of the background contribution as is practical. If a roughly 5% or 10% variation in the background contribution remains in these data, which seems plausible, then that is a contribution to the variation of the results that will be unavoidably present in the images.1214.4.3.2 ICV InjectionsIn the animals that received ICV targeted injections, the residual background after constant subtraction required a different correction technique than that discussed above due to their distinctive distribution in relation to the tissue slices. Background was notably seen to decrease with distance from tissue slices, and was generally larger in regions with more total activity in nearby tissue slices. An example of background between slices is seen in figure 4.11.Figure 4.11: Section of autoradiograph of rat that received ICV injection of Mn-52, centred between4 brain tissue slices (one at each corner). This image has been rescaled to show the spatiallyvarying background after constant subtraction from the plate. Background outside of the tissuesections decreases with distance from the tissue, suggesting a similar pattern will continues withinthe tissue regions, making a simple background subtraction inadequate.To correct this shape of background, the previously-discussed method of a separate but constant background subtraction for each slice was judged insufficient, due to the variation of the background within the slice region on a spatial scale similar to the slices. Instead, a second-step background correction was implemented by estimating and subtracting a spatially-varying background distribution for entire autoradiographic images, prior to separating into single-slice cropped regions.122Because the background appeared to depend on the amount of activity in the nearby tissue regions, a convolution method was used to estimate the background shape. The convolution method involved first creating a thresholded version of the activity distribution after constant background subtract. This was done similarly to the background isolation for the IP injected image, except that inthis case only activity above the threshold was included, keeping the tissue regions and excluding the surrounding background regions. The thresholded image was used as an estimate of the source distribution of the activity that was leading to the background distribution, and was convolved with an kernel in order to reproduce the observed shape of the background. The kernel was produced iteratively by circles of various sizes and weights to produce a cylindrically symmetric pyramidal step function, convolving with the thresholded activity distribution and comparing with the background and its the falloff with distance at the edges of the grid of tissue slices. An example background generated in this manner is shown in figures 4.12 and 4.13.Figure 4.12: Estimated background distribution after convolving kernelwith image intensity within tissue slice regions. Arbitrary scale.Figure 4.13: Estimated background distribution masked with tissueslice locations, and with a representation of the convolution kernel at centre.123Background distribution estimation was performed on the constant-background adjusted autoradiographs after down-sampling using bicubic interpolation with anti-aliasing, to increase the size of pixels in both axes by a factor of 25. On the down-sampled image, the kernel had a total width of 60 pixels. A visual representation of the kernel appears in the centre of figure 4.13. The kernel consisted of summed concentric circles with sizes relative to the kernel radius and relative intensities of: radius/1.0 with intensity 0.9, radius/1.5 with an additional 0.25 intensity, radius/3.0 with an additional 2.5 intensity, radius/4.5 with an additional 2.0 intensity. After convolving the kernel with the thresholded image, the background was rescaled manually to as best as possible match the magnitude of the background seen in the full autoradiographs after constant background subtraction. A single scaling factor was used for the whole of each autoradiograph, rather than a separate scaling factor for each plate. (As noted above, the full autoradiograph contains two separate autoradiographic plate images, shown at left and right in the above figures, separated by a gap with no slices at the centre of the image.) The background was then subtracted from the autoradiographs.A two-step background distribution estimation was briefly tested, in which the initial background distribution was estimated as described above, was subtracted from the image, and then a second estimate was generated from the corrected data using the same convolution method. The resulting second-step distribution estimate was nearly indistinguishable from the single-step result, and was thus judged unnecessary.As with the corrections for the IP injected animal, these background corrections are not expected to produce perfect results. Rather, then are intended to remove as much of the background variation as possible. Due to the variation in the background in these images, however, it is possible to examine the full images after background corrections to observe the scale of any remaining background in the images from the remaining deviations of the background pixel values from the ideal of 0 background. Corrected full autoradiographs are shown at scales to illustrate the remaining background in figures 4.14 and 4.15.124Figure 4.14: Full autoradiograph after background corrections have been applied.Colour scale is set to show remaining background variations after corrections.Figure 4.15: Full autoradiograph after background corrections have been applied.Colour scale is set to show activity distributions within slices.As seen in the figures above, the scale of the residual background variation after correction equal or less than less than +/-200 digital light units (DLU), while the intensity of the images in the regions of tissue slices ranges from 2000 DLU to over 5000 DLU (and actually reaches over 104 DLU in the peak regions, beyond the scale of the figure here). As such, the residual background level is below 10% formost of the tissue pixels. While this background and its variation will introduce some variation in results when comparing with other modalities, this level of accuracy was deemed adequate for the comparisons in this work. Spatially varying or more rigorously optimized kernel shape and scaling could be employed to further improve background removal if necessary for future work.1254.5 Slice RegistrationBecause autoradiography images single ex vivo tissue slices, it does not inherently provide a volumetric intensity distribution as is provided by PET and MRI. In order to compare between AR and PET or MRI, it was necessary to combine the 2D slice images into an approximate 3D volume representing the brain tissue from which the slices were cut. This was only practical for the rats whose brains were sliced perpendicular to the superior-inferior axis, as those brains had closely spaced slices that covered a large portion of the brain volume. For the two rats whose brains were sliced perpendicular to the anterior-posterior axis, the spacing between slices was quite large, and did not cover enough of the brain volume to make combining the slices into a volume practical or useful.Autoradiography slice registration was done using a custom written MatLab script. Areas on the autoradiographic images that enclosed each tissue slice's image were specified, and stored as separate cropped single-slice images. Numerous examples of these single slice images and the full autoradiographic images from which they were cropped are shown in the previous sections of this chapter.After extracting all tissue slices, it was necessary to apply geometrical transformations to make the orientation of the slices consistent. For example in the full autoradiograph images shown above of slices cut perpendicularly to the inferior-superior axis of the brain, roughly half the plates have the posterior end of the brain at the top end of the brain slices in the images, while the other half have that end at the bottom end of the brain. Similar orientation variation were seen in the slices cut perpendicularly to the anterior-posterior axis. To correct this variation, slices were rotated through 180 degrees when necessary so that the posterior or inferior ends of the slices were at the bottom of single slice images. It was clear that 180 degree rotations were needed, rather than mirroring of the images in the up-down direction, due to asymmetries in the distributions of intensities of the images, as can be seen in some of the slice images above.Because the slices were placed on the imaging plates by hand, they also had small angular variationsin their orientations even after 180 degree rotations. As well, the cropping procedure to isolate 126individual slice images was imprecise, and there were translational variations of the positions of the images of slices within the cropped regions, and thus within the single slice images. Correcting thesevariations required an additional coregistration step, in which each slice was translated and rotated using MatLab image rotation functions to better match the overall position of slices in the brain, and particularly to match the positions of its neighbour slices. This was done using custom-written interactive scripts that displayed the slices and the differences between slices and their neighbours, and allowed the slices to be reoriented. A calculated cost function was not used, for reasons discussed below; instead, visual inspection was used to guide the registration.There were practical limits to the effectiveness of the procedure, however, due to distortions and damage to tissue slices during the slicing process. Examples of such damage is shown in figures 4.16,4.17, and 4.18.Figure: 4.16: AR slices from brain of rat that received IP injection of Mn-52 that show largeregions of damage. These slices were excluded from the volume for this animal,and were replaced with adjacent slices.When creating the volume image from AR slices for the IP injected animal, two severely damaged slices were replaced with data from adjacent slices, as seen in figure 4.16. The omitted slices were adjacent to each other, and so the other adjacent but non-(or less-)damaged slice for each was used as the replacement, rather than an average of adjacent slices which would have included one of the damaged slices. In most other slices, less severe damage could be seen, such as in figures 4.17 and 4.18.127Figure 4.17: AR slice from brain of rat that received misplaced ICV injection of Mn-52 that showsminor damage and overlap of tissues. At the top of the slice, the tissue has a concave shapedue to damage or lost tissue. At the bottom of the slice, another slice is partially overlapping. At thebottom-right, there is a streak of high activity at the edge of the slice, possibly due to overlap ofthe slice with itself. This level of damage was insufficient to exclude a slice from the tissue volume for thisanimal because most of the brain is unaffected, and the general shape of the distribution is intact.Figure 4.18: AR slices from brain of rat that received successful ICV injection of Mn-52. In the leftimage, a chunk of cortex is missing. In the right image, at the top and bottom, regions of increasedbrightness relative to the surrounding tissue, along with the flat border with the background regionsuggest that small section of tissue have folded over adjacent tissue, effectively combining theiractivities in the overlapping region and leaving the original location of the folded-over tissue as background.These levels of damage were insufficient to exclude slices from the brain volume for this animal.No slices were replaced when creating volume AR images of ICV injected animals, because the majority of the slice areas were intact and provided potentially useful information about the Mn-52 distribution. The damaged regions were also generally on the periphery of the tissue regions in the slices, so that the interesting details of the distribution in the middle of the slices was visible.128Registration was guided, and its quality judged, by comparing slice positions visually, and adjusting the registration transformation until it was as optimized as reasonably possible. In part due to the aforementioned tissue damage, no mathematically-calculated metric of registration quality was usedto guide registration of tissue slices while generated the 3D volume images. Additionally, manually registration is still often used as the gold standard for image registration, and a major benefit of automated methods is saving investigator time. For the relatively small number of images being registered in this work, using manual registration was not problematically time consuming.Even with this manual registration, however, the result was not a perfect replication of the 3D brain activity distribution. The process of freezing and partially thawing the rat, extracting the brain, and slicing it onto slides unavoidably introduces some spatial deformations to the tissues (in addition to the readily visible damaged sections discussed above). After registration, the AR image volumes were saved as single 3 dimensional image files for each animal.1294.6 ResultsSelected slices from autoradiography images are shown in figures 4.19 through 4.23 below, after all corrections have been applied.Figure 4.19: Autoradiography slice cut perpendicular to the superior-inferior axis after a successfulinjection of Mn-52 into the right lateral ventricle. The injection site appears as a distinct asymmetrical feature,marked by the white arrow. The colliculus appears as a symmetrical large accumulation, left of the black arrow,with the colliculus at its centre. The cerebellum appears with a moderate accumulation of activity, marked bythe grey arrow. The cortex has a lower level of activity, surrounding the sides and front of the brain in the image.Figure 4.20: Autoradiography slice cut perpendicular to the superior-inferior axis after amisplaced injection of Mn-52 into brain tissue near the right lateral ventricle. Activity appearsin a V-shaped pattern near the brain midline, marked with the white arrow, possibly due totransport across the corpus collosum from the injection site. The colliculus appears, markedwith the black arrow, as a region of higher uptake than the surrounding tissue.130Figure 4.21: Autoradiography slice cut perpendicular to the superior-inferior axis after an IPinjection of Mn-52. Activity appears in a similar pattern to the ICV injection, with the colliculus,marked with the black arrow, and cerebellum, marked with the grey arrow, brighter thanthe cortex, but without any bright spot in the lateral ventricles.Figure 4.22: Autoradiography slices in a rat that lived for approximately three hours after ICVinjection of Mn-52. These slices was cut perpendicular to the anterior-posterior axis afterinjection of Mn-52 into the right lateral ventricle. At this short time after injection, the distributionof activity in the brain is likely dominated by the positions and sizes of the intracerebral ventricles,through which Mn could spread without needing to cross a brain barrier.131Figure 4.23: Autoradiography slices in a rat that lived for approximately one day after ICV injection of Mn-52.These slices was cut perpendicular to the anterior-posterior axis after injection of Mn-52 into the right lateral ventricle.These autoradiographs provide detailed images of Mn-52 distribution in rat brains that will be usefulfor comparison with images acquired with other modalities in this work. Particularly for comparison with MR images, these images provide much better ability to resolve details of the distribution than do PET images shown in the PET Imaging chapter.1325 Magnetic Resonance Imaging BackgroundNuclear magnetic resonance imaging (MRI) is a medical imaging technique with the ability to reveal anatomical and functional information through contrasts between soft tissues within living subjects. There are many variations of MRI, which can probe different and complementary information about the subject. The variations of MRI also make it very adaptable, for small animals like mice or rats, ex vivo tissues, or body human imaging in various anatomical positions.This section discusses the source of MRI images, and selected topics related to how they are acquired and analyzed.5.1 Physical BasisMagnetic resonance (MR) produces images by measuring the radio-frequency (RF) oscillating magnetic field from an ensemble of atomic nuclear spins precessing and relaxing towards thermal equilibrium in a constant externally-applied magnetic field. This relaxation occurs after spins are excited out of equilibrium by the MR system applying an additional external RF field oscillating at thespin precession frequency. Image spatial information is encoded in phase and frequency of the precession using magnetic field gradients.The following text discusses the origin of the MR signal from the difference in energies of spin states in an applied field in a large population of spins.5.1.1 Nuclear Spins and Macroscopic MagnetizationAtoms that can be measured with magnetic resonance experiments are those with nonzero nuclear spin. These are isotopes in which the number or protons, the number of neutrons, or both these numbers, are odd. If both numbers are even, the spins of these nucleons will arrange in pairs in opposite orientations in the minimal energy available states, giving a net zero nuclear spin. If one or both numbers are odd, the nucleus will have a net spin, and will be able to produce a measurable 133signal when placed in a magnetic field.The are many isotopes with nonzero nuclear spin, including F-19, Na-23, and P-31. The most important isotope for MRI is H-1, a single proton. Hydrogen nuclei are particularly useful for medical and preclinical research because large amounts of measurable hydrogen nuclei are present in biological tissues due to their water and fat content. Subsequent discussion will focus on hydrogen MRI, although similar principles apply to other nuclei used in MR experiments.Protons are spin 1/2 particles, with two possible results that may be observed if their spin is measured in a particular direction, corresponding to the eigenstates of the spin measurement operator, which are labelled with the quantum number m with values of +1/2 or -1/2. When there is no external magnetic field, these spin states are degenerate, and a measurement of the spin state in a proton from a population at thermal equilibrium will have equal probability of observing either result.When an external magnetic field is applied to a spin 1/2 system, Zeeman coupling between the spin and the applied field breaks the degeneracy of the spin states. The m = +1/2 eigenstate of the spin measurement in the direction aligned with the applied field will have slightly lower energy than the m = -1/2 eigenstate, due to this coupling. The strength of the coupling between the applied field, B0, and the spin states is described by the gyromagnetic ratio ? and the reduced Planck's constant,? = 4.14x10-15 eVs: ?E=??B0The value of ? depends on the particle, and for H-1 nuclei is approximately 42.57 MHz/T. For an applied field of 7 T, this corresponds to an energy difference of 1.23 ?eV.Because practical MR experiments operate on objects containing large numbers (on the order of Avogadro's number) of spins, it useful to consider an ensemble of such spin states being measured simultaneously. At thermal equilibrium, due to the energy difference between states, there will be a slightly higher probability of measuring any given spin in the lower energy state. Each spin may independently be measured in either of the spin states, leading to a statistical distribution of state 134occupancy. In thermal equilibrium, the results of an observation of the occupied states will be described by a ratio of Boltzmann distribution values:N ?N + =e??Ek BT ?1? ?Ek BTWhere N+ and N- are the numbers of spins in the m = +1/2 and m = -1/2 states, respectively, andkB = 8.62x10-5 eV/K is Boltzmann's constant. At 37 C or 310 K, kBT is 26.7 meV, and the ratio of state occupancies is approximately 0.999954. For a population of 106 spins, the difference in expected numbers of protons measured in the two states is N+ - N- = 23. Because of the large number of proton spins in water or biological tissue, this relatively small difference in occupancy produces a measurable macroscopic magnetization in objects in MR experiments.The energy difference between spin states in an applied magnetic field has several other physical interpretations. As noted for protons at 7T, the energy difference is 1.23 ?eV or 1.97x10-25 J. Transitions of protons between these spin states will involve interactions with photons of this energy; absorption of a photon can cause a spin to transition from the lower-energy to the higher-energy state, and the reverse transition will involve an emission of a photon to conserve energy. These emitted photons also have a corresponding electromagnetic frequency, which in the case of protons at 7T is approximately 297.4 MHz, which is in the radiofrequency range. This frequency is also the classical rate of precession of the ensemble magnetic moment in an applied field when not at thermal equilibrium, which produces a time-varying electromagnetic field with that same frequency. The angular frequency of this precession ? is known as the Larmor frequency, and is given by the relationship:?=?B0where ? is again the gyromagnetic ratio for the particle producing the magnetic moment, and B0 the applied external field. Because MR systems do not measure individual spin states, but rather the combined behaviour of an ensemble of spins, it is this macroscopic RF field interpretation that is most useful for describing MR experiments. 1355.1.2 Signal ReceptionTime-varying magnetization in an MR experiment produces a time-varying magnetic field, which oscillates at the Larmor frequency. This field is the classical source of the signal measured by an MRI scanner.The sensitivity of any reception coil design to precessing magnetization at some location is proportional to the efficiency of that same coil being used as a transmitter to generate a radiofrequency magnetic field at the same location. The reciprocity theorem for antennas indicates that the vector-valued sensitivity profile BRef(r) of a coil at a location, r, is equal to the magnetic field at that location that would be produced by a unit current passing through the coil. For a time-dependent distribution of magnetization M(r), the voltage in the coil is determined by the volume integral of the dot product of the magnetization and the sensitivity profile (Buonocore, 2013; Tropp, 2013):V (t )=? ddt? M? ( r? ' )?B?Ref ( r? ' )d 3 r? 'For an MR experiment, only the components of the magnetization in the plane perpendicular to the main applied field will precess, and the generated RF electromagnetic field components will be oriented in this plane as well. The strongest signal from an MR reception coil will be measured when the sensitivity profile of the coil is in the plane of precession of local magnetizations. For coil geometries similar to a planar conducting wire loop, the coil would optimally be oriented such that its surface is near parallel to the applied field, or equivalently, so that the normal to its surface is in the plane in which magnetizations will precess (as in figure 5.1). In practice, other coil designs and experimental considerations may motivation use of different configurations.136Figure 5.1: Precession of magnetization, M, in XY plane, perpendicular to external applied magneticfield, B0, that is aligned along the Z axis. Single turn signal reception coil is shown oriented with its surfaceparallel to the external applied field to detect oscillating signal from magnetization precession.5.1.3 PrecessionAs noted above, macroscopic magnetization in an applied magnetic field will precess around the direction of that field, unless the magnetization is aligned with the applied field, in which case there will be no macroscopic precession. An arbitrary macroscopic magnetization may be described by a 3D vector, or by its components in orthogonal basis directions:M?=(M xM yM z)The z-direction magnetization component, Mz, is generally assigned to be aligned with the applied magnetic field, and is often referred to as the longitudinal magnetization. The x and y components, Mx and My, lie in a plane perpendicular to the applied field, and are often referred to as the transverse magnetization.The time-rate of change of the magnetization due to precession in an externally applied magnetic field may be written:ddt M?=? M?? B?137If, by convention, the vector B is nonzero only in the z-direction, due to the static applied field, labelled B0:ddt (M xM yM z)=( 0 ?B0 0??B0 0 00 0 0)(M xM yM z)where as above the rate constant is the Larmor frequency for the spins in the applied field:?0=?B0For an initial magnetization M0 = (Mx0, My0, Mz0), this vector equation has solution:(M x(t )M y (t )M z (t))=(M x0 cos (?0 t )?M y0 sin (?0 t )M x0 sin(?0 t)+M y0 cos(?0 t )M z0 )These equations may also be rewritten in frame of reference that rotates with the same angular frequency, ?0, as the magnetization. This transformation fixes the orientation of the precessing transverse magnetization. Equivalently, in this reference frame, there is an effective longitudinal magnetic field of zero magnitude, B0' = 0:(M x ' (t )M y ' (t)M z ' (t ))=(M x0 'M y0 'M z0 )where ' indicates quantities in the rotating reference frame, and the fixed and rotating reference frames coincide at the time when the initial magnetizations are defined. This transformation is particularly useful for simplifying the expressions for the changing magnetization when additional terms are introduced to the matrix differential equation. In particular, the above result neglects the important effects of relaxation, and terms describing those effects on magnetization may be added, as discussed below.1385.1.4 RelaxationThere are two types of relaxation relevant to magnetic resonance experiments: spin-lattice, and spin-spin interactions. These relaxation mechanisms are important for MRI, as differences in the rates of relaxation lead to contrast in MR images.Spin-lattice relaxation is the process by which the longitudinal magnetization, in the direction of the applied field, approaches its thermal equilibrium value. Particles in excited states may transition to lower-energy states by emission of a photon, however such transitions occur relatively slowly without stimulated emission. For radiofrequency energies, as occur in magnetic resonance experiments, fluctuations of the magnetic fields to which a spin is exposed may stimulate such a transition.Variations in magnetic fields that lead to relaxation arise because magnetic fields in a material are not uniform in time or space. Molecular motion, rotation, exchange of atoms between molecules, and variation of the local composition of materials at these scales produce spatially and temporally varying fields, to which spins are exposed.Not any variation in local fields will effectively induce longitudinal relaxation, however. The Larmor frequency at which spins precess also determines the frequency of changing magnetic field that can most effectively induce transitions between spin states. The temporal frequency of magnetic field variations to which a spin is exposed will depend largely on the durations of interactions between the spin and local spatial field variations.An important contribution to the duration of field variations is the 'tumbling rate' for the molecules containing the excited spins. This is the rate at which the molecule rotates and translates nearer and further from sources of field variation. The tumbling rate depends on a variety of factors, including the size of molecules involved, temperature, viscosity, physical state of matter, biological micro-structures, and what other molecules are present in the material. Molecules with tumbling rates similar to the Larmor frequency will experience variations in magnetic fields at the same rate, and will thus see faster relaxation of their longitudinal magnetization. At higher or lower tumbling rates, 139longitudinal relaxation will be slower.Strength of the applied magnetic field also influences the rate of relaxation. For the same molecule, with the same tumbling rate, exposed to different applied field strengths, the Larmor frequency will be different. The change in the Larmor frequency with fixed tumbling rate may bring these two rates closer together, speeding up relaxation, or further apart, slowing the relaxation process.Regardless of the mechanisms discussed above, longitudinal magnetization for a population of spins is often approximately described as having an exponential decay or relaxation, from its initial value Mz0 towards its thermal equilibrium value, Meq:M z(t )=M eq?(M eq?M z0)e? tT1The characteristic time for this relaxation is T1. Alternatively, the relaxation rate R1 may be used instead, where R1 is 1/T1.This relaxation with time may also be written in the form of a differential equation for Mz:ddt M z (t )=M eq?M z0T1 e? tT1=?M z(t )+M eqT1This expression, and the previous explicit time dependence of Mz are unaffected by transformation into a rotating reference frame, and have the same form in that frame.Longitudinal relaxation may also be added to the vector equation for the magnetization rate of change:ddt (M xM yM z)=( 0 ?B0 0??B0 0 00 0 ? 1T1)(M xM yM z)+( 00M eqT1 )140The T1 relaxation time is a useful parameter to characterize magnetization relaxation in MR experiments because different tissues and tissue environments will relax at different rates. This permits the T1 relaxation time of a sample to be used to characterize that sample, as is further discussed below.The second primary type of relaxation in magnetic resonance experiments is spin-spin relaxation. This is the process by which the transverse magnetization, perpendicular to the applied field, decays to zero. Like spin-lattice relaxation, local magnetic field variations also lead to spin-spin relaxation, but through a slightly different mechanism. Because the component spins that collectively create the macroscopic magnetization are exposed to spatially- and temporally-varying magnetic fields, the precession frequency of those component spins will not have a single fixed value as implicitly assumed above. Rather, the macroscopic magnetization will have a random and relatively narrow distribution precession rates, which leads to a spread of phase of precession with time. Macroscopic magnetization is maximized when all the spins contributing to it have the maximum degree of phase coherence, so dephasing of spins results in a reduction of the transverse magnetization with time.The rate of transverse dephasing or spin-spin relaxation depends in part on the molecular tumbling rate, as is the case for the longitudinal or spin-lattice relaxation. Unlike the spin-lattice relaxation, there is no local maximum of the spin-spin relaxation at a particular tumbling rate. Instead, for spin-spin relaxation there is a consistent trend for higher tumbling rates to lower the rate of relaxation. This occurs because the spin dephasing involves an accumulation of phase differences, and at slowertumbling rates, spins are exposed to local variations in the magnetic field for longer times, and can thus accumulate more phase difference from the ensemble average phase. As tumbling rate increases, the variations in phase from transient field strength variations are averaged in time, leading to a net lower accumulation of phase spread for spins.Regardless of the mechanisms discussed above, transverse magnetization for a population of spins isoften approximately described as having an exponential decay, from its initial value Mxy0 towards its thermal equilibrium value of 0.141In the rotating frame:M xy ' (t )=M xy0 e? tT2ddt M xy ' (t)=?1T2 M xy ' (t)The characteristic time for this decay is T2 or T2*. Alternatively, the relaxation rates R2 or R2* may be used instead, where R2 is 1/T2 and R2* is 1/T2*.  The distinction between R2 and R2* is discussed below.In the non-rotating frame:ddt (M xM yM z)=(? 1T2 ?B0 0??B0 ? 1T2 00 0 ? 1T1)(M xM yM z)+( 00M eqT1 )As with T1, the T2 relaxation time is a useful parameter because different tissues and tissue environments will relax at different rates in MR experiments. This permits the T2 relaxation time of asample to be used to characterize that sample, as is further discussed below.A portion of transverse magnetization dephasing (and thus relaxation) occurs due to components of the magnetic field spatial distribution that vary relatively slowly with position. For these contributions to dephasing, the tumbling of molecules does not substantially alter the field experienced by proton spins. Stated another way, these dephasing contributions are not randomly-varying with time, but rather are fixed in the neighbourhood of each spin location.This has an important consequence for MR imaging, in that the fixed variations in magnetic fields remain fixed even if the spins are manipulated by radiofrequency pulses, as discussed below. This permits this contribution to spin dephasing to be effectively removed from the decay of signal magnitude.142The total rate of transverse magnetization magnitude decrease is characterized by the rate parameter T2*, or its inverse R2*. The rate parameters T2 or R2 are reserved for the contributions tothe transverse relaxation that cannot be corrected with radiofrequency manipulation of the spin system. Alternatively, T2 includes only irreversible dephasing of the transverse magnetization, while T2* includes reversible dephasing.The times taken for these relaxation effects are typically on the order of ms to s in biological tissues, but vary greatly depending on the particular tissue or material, applied field and details of an imaging experiment. These distinct relaxation effects can be separated and characterized by appropriate experiments using an MRI system, as discussed below.5.2 Radiofrequency PulsesMacroscopic magnetization in biological tissues, placed in a magnetic field of the strength used for MRI (on the order of 1 to 10 T), will generally reach effective thermal equilibrium within 1 to 10 seconds. In order to acquire data for MRI, it is necessary to excite the system from equilibrium, into a more energetic state where the magnetization has a component in the transverse plane, and where it will precess and generate a measurable radiofrequency (RF) signal.The tool used to excite spins in this manner is to expose the object to an applied RF electromagnetic field, with electrical and magnetic field components in the transverse plane. Controlled duration andstrength ?pulses? of RF field of this nature can be used to manipulate the magnetization in various ways to prepare an object for an imaging experiment. There are an innumerable variety of possible RF pulses that may be used in this manner, with various advantages and drawbacks such as the RF power deposition or frequency bandwidth inherent in limited duration pulses, which may affect the magnetization manipulation results.1435.2.1 Physical DescriptionIf the frequency, ?1, of an oscillating applied external RF magnetic field, B1, matches the Larmor frequency, ?0, of spins in the fixed applied external magnetic field, B0, a resonance will occur between the oscillating field and the spins in an object. This resonance is essentially the same mechanism by which longitudinal relaxation occurs, when the molecular tumbling rate is matched tothe Larmor frequency, as discussed above.In the quantum description, this resonance implies that the energy of photons in the RF field is matched to the transition energy between the spin states of the nuclei being imaged, which allows those photons to stimulate transitions between spin states.In the macroscopic description, this resonance implies that a precessing magnetization will have a consistent phase with respect to the B1 field. The presence of this B1 field has the effect of changing the effective direction of the magnetic field about which the macroscopic magnetization precesses, as illustrated in figure 5.2. Rather than being aligned with main applied B0 field, the vector sum of B0 and B1 will be an effective field, Beff, oriented somewhere between the longitudinal axis and transverse plane, depending on the relative strengths of the B0 and B1 fields.Figure 5.2: Instantaneous precession of magnetization, M, about the effective applied magnetic field, Beff,that is the sum of the static main external applied field, B0, and the rotating RF field, B1.144As illustrated in figure 5.3, while the B1 field is active, the magnetization will rotate away from its initial direction, towards the transverse plane, at an angle in the transverse plane 90 degrees from the angle of the B1 magnetic field in the rotating reference frame. By controlling the duration and strength of the B1 field, the magnetization orientation can be rotated to any desired angle with respect to the main applied field direction.Figure 5.3: Excitation of magnetization, M, by rotation away from the direction of the mainexternal applied field, B0. Shown in the rotating frame where the magnetization componentin the X'Y' plane (the component perpendicular to the main applied field in the Z direction)does not precess about the Z axis. Excitation occurs due to precession of the magnetizationabout the RF field, B1, here shown oriented along the X' axis of the rotating frame.The B1 field effect on the magnetization rate of change may also be added to the full fixed reference frame vector differential equation:ddt (M xM yM z)=(? 1T2 ?B z ??B y??B z ? 1T2 ?Bx?B y ??B x ? 1T1 )(M xM yM z)+( 00M eqT1 )where the B0 field is here represented as Bz, and the B1 field appears as the Bx and By components. This reproduces the Bloch equations for spin dynamics, incorporating precession, transverse and 145longitudinal relaxation, and excitation due to an applied RF field. This coupled set of rate equations describes the basic framework of magnetic resonance spin behaviour and manipulation.5.2.2 Signal FormationSeveral types of RF pulse are particularly notable as the basic tools by which MR experiments manipulate spins to generate MR images. A discussion of some of these pulses and their applicationsappears below, with emphasis on different contrasts that can be generated between tissue types. Later, several specific types of MRI experiment that are directly relevant to the experiments in this work are discussed in greater detail.5.2.2.1 90 Degree PulseThe conceptually simplest MR RF pulse is the 90 degree excitation pulse. When applied to a fully relaxed magnetization system, a 90 degree RF pulse will rotate magnetization from the longitudinal or main field direction, into the transverse plane. This produces the largest possible precessing magnetization for this spin system, as larger or smaller rotation angles will leave less spin magnitude in the transverse plane. The magnitude of the signal immediately after excitation will depend most notably on the density of proton spins in the object. This variation in signal leads to proton-density weighted contrast in images.5.2.2.2 T2* ContrastAfter magnetization is rotated into the transverse plane, it will begin to relax and dephase as described above. In the rotating reference frame:M xy ' (t )=M xy0 e? tT2?This loss of net transverse magnetization, as illustrated in figure 5.4, leads to a reduction of measured signal with time, characterized by the T2* relaxation time or the R2* relaxation rate, as the signal is proportional to the precessing magnetization magnitude. If measured some time after 146the excitation, the MR signal will have decreased by an amount depending on the delay time and theobject's T2* relaxation time. If the delay time is in a range comparable to the T2* times of tissues, different tissues will exhibit measurably different signal magnitudes. This variation in signal leads to T2*-weighted contrast in images.Figure 5.4: Exponential decay of transverse magnetization with two characteristic times.Different characteristic times produce different remaining magnetizations with time,leading to contrast in images of objects containing materials with different times.5.2.2.3 T1 ContrastAfter an excitation pulse is applied, the longitudinal magnetization component begins to recover or relax towards its thermal equilibrium value, as illustrated in figure 5.5, at a rate characterized by the T1 relaxation time or the R1 relaxation rate. After a 90 degree excitation:M z(t )=M eq(1?e? tT1 )During this recovery, if a second 90 degree excitation pulse is applied to the magnetization, whateveramount of longitudinal magnetization present at that time will be rotated into the transverse plane. The magnitude of the resulting RF signal will be proportional to the magnitude of the longitudinal magnetization at the time the excitation was applied. Unless a sufficiently long time elapsed between the excitations for the longitudinal magnetization to fully recover to its thermal equilibrium147value (often approximated as within 5 x T1 after the previous excitation), the signal magnitude after the second excitation will be less than its value when the excitation is applied at thermal equilibrium. Different objects or tissues have different T1 times, so if the period between excitations is suitably short, the different tissues will exhibit different signal magnitudes. This variation in signal leads to T1-weighted contrast in images.Figure 5.5: Exponential recovery of longitudinal magnetization with two characteristic times.Different characteristic times produce different magnetization recovery with time,leading to contrast in images of objects containing materials with different times.5.2.2.4 180 Degree InversionAfter the 90 degree pulse, the next most fundamental MRI RF pulse is an inversion, "flip", or 180 degree rotation of the magnetization. As with the 90 degree pulse, the axis about which this rotationoccurs lines in the transverse plane. The effects of an inversion can be described by separately considering how it manipulates longitudinal and transverse magnetization components. For magnetization components in the transverse plane, 180 degree rotations will have the effect of mirroring the magnetization across the transverse plane. Notably, this mirroring will only affect the components of the transverse magnetization that are perpendicular to the line through the origin across which the mirroring occurred. A magnetization, or component thereof, that lies along the mirroring line will be unaffected, as illustrated in figure 5.6.148Figure 5.6: Magnetization vectors in the transverse plane before (left) and after (right) application ofa 180 degree inversion pulse that rotated the magnetizations about the (vertical) y-axis.For magnetization components in the longitudinal direction, 180 degree rotations will have a similar effect of mirroring. Because the inversion mirroring line lies in the transverse plane, longitudinal components will always be inverted by such a pulse.Inverting or mirroring magnetizations in this manner is an essential part of numerous MRI pulse sequences. Two of the more fundamental of these sequence types, spin echo and inversion recovery, are discussed below.5.2.2.5 T2 Weighting / Spin EchoIf a magnetization has been rotated into the transverse plane and is precessing, it will undergo T2* relaxation, losing magnitude as magnetizations within a sampling volume precess at different rates and lose phase coherence. As discussed above, the T2* magnetization loss includes both reversible and irreversible components, and it is often desirable to minimize the loss of transverse magnetization, and thus signal, by reversing as much of this loss as possible. This reversal will leave only the irreversible T2 component of the magnetization loss. Application of 180 degree refocusing pulses is the primary tool used to accomplish this.In the reference frame rotating at the Larmor frequency, after excitation into the transverse plane, the magnetization is ideally held at a fixed angle. In practice, precession rate variation leads to a progressive accumulation of phase with time, so that magnetization components within an imaging 149volume acquire a spread of phases, as illustrated in figure 5.7. The reduced coherence of these magnetization components causes their sum to progressively decrease in magnitude.Figure 5.7: Magnetizations all oriented in the same direction of transverse plane immediatelyafter excitation (left), all with the same phase. After some time precessing at slightly differentrates (right), magnetizations have spread out in phase. Shown in a rotating reference frame.If a 180 degree inversion is applied about the axis (the horizontal x-axis in figures 5.7 and 5.8) aligned with the magnetization precessing at precisely the Larmor frequency, the magnetizations willbe mirrored across that axis, as illustrated in figure 5.8. One interpretation of this transformation is that the relative phase of each magnetization component is negated. That is, magnetizations that precessed slightly faster than average will have accumulated a positive phase offset from the average, and magnetizations that precessed slightly slower will have accumulated a negative phase offset. The inversion will reverse these relative phases, so that magnetizations that were precessing faster have a negative phase offset, and the converse for the slower-precessing magnetizations. The inversion does not, however, change rate of precession of these magnetizations. As such, the phase offsets after the inversion are inversely proportional the differences in rates of precession after the inversion.As a consequence of the inversion, the relative offsets if phase of magnetizations will begin to decrease. After a time equal to the delay between the initial excitation and the inversion, the magnetizations will have accumulated approximately equal phase as they did before the inversion, and their overall phases will return to nearly equal, as illustrated in figure 5.8.150Figure 5.8: Magnetizations after phase spread during precession and inversion about the x-axis (left).After another period of precession, magnetizations have rephased to nearly the same direction, althoughsome differences may remain due to irreversible contributions to precession rate variations.Rephasing of magnetizations in this manner will generate an increase in signal magnitude with time, as the sum of magnetizations increases as the spread in phase decreases. After a time equal to twicethe delay between the initial excitation and the refocusing, a peak magnetization will occur, which is often referred to as the "spin echo". The magnitude of the magnetization, and thus signal, at this time will depend on the irreversible T2 contribution to the transverse magnetization decay, as the additional reversible contributions have been cancelled out in creating the echo. The time at which the echo occurs is often referred to as the echo time, labelled TE.M xy ' (TE )=M xy0 e?TET2Additional inversion pulses may be applied to produce additional echoes, allowing multiple measurements of the spin magnitude after a single excitation pulse. These additional echoes may begenerated at regular times, equal to multiples of the first TE, or may be produced at arbitrary timing,by adjusting the delay between subsequent inversion pulses. In any case, the signal magnitude at the subsequent echo times will be determined by T2, rather than T2*.Alternatively, spatially-selective (see below) refocusing may be applied after a spatially-selective excitation. Spatial selectivity is often applied in a single direction, so that a thin slice of an object is affected, while the remainder is not resonant with the applied RF pulse. If the spatial limits of a series of RF pulses are oriented in orthogonal directions (labelled, z, x, and y), an excitation localized in z, followed by inversions on x and y, only a small volume at the intersection of the three planes 151will have spin echoes generated with it. By this means, a limited volume region may be isolated within an object as the sole source of signal. This may be done without the phase and frequency-based spatial localization discussed below, and is used for localized spectroscopic measurements in MR scanners.5.2.2.6 Inversion RecoveryWhen an inversion pulse is applied to a fully relaxed magnetization system, the magnetization is rotated from fully aligned with the main applied field to a direction fully opposing that field. This produces no measurable signal, as there is no component of magnetization placed in the transverse plane. As was discussed above regarding T1 contrast arising from delays between subsequent excitations, after an inversion, the longitudinal magnetization will relax towards its equilibrium value. As well, if a 90 degree excitation pulse is applied during this relaxation, the resulting signal magnitude will depend largely on how much time passed between the inversion and excitation. This time, often referred to as the inversion time, is also labelled TI:M z(TI )=M eq(1?2 e?TIT1 )Two applications for this magnetization manipulation are particularly notable.First, if an object has magnetizations within it that have different longitudinal relaxation rates, such as fat and water within biological tissues, an inversion recovery pulse sequence can be used to isolate these signal contributions. Due to the different T1 times of these tissue magnetizations, the times between inversion and the moments when the two tissues will have zero longitudinal magnetization is different. If it is desired to produce an image that excludes one but still shows the distribution of these tissues, the inversion delay (TI) may be chosen coincide with the time when the excluded tissue has zero longitudinal magnetization. The zero-magnetization tissue is often referred to as "nulled" or "suppressed" at this time. A 90 degree excitation pulse applied at that time will result in a signal that arises only from the non-nulled tissue.152Second, it is sometimes useful to determine the numerical value of the T1 (or T2) that characterizes the relaxation rate of a tissue. This parameter can be related to various changes in the tissue, some of which are discussed below. In order to do so, it is insufficient to have a single T1-weighted image, as the relationship between a tissue's T1 and the signal magnitude at a particular time has numerous parameters that are not known. Several methods for measuring the T1 of objects are discussed in a separate section below (see page 158).5.2.2.7 Fast Repetition TimeIn addition to the aforementioned 90 and 180 rotations, any angle may be chosen when applying RF pulses to rotate magnetizations in MR experiments. Other angles have a somewhat more complicated effect on the magnetizations, however.In the case of a magnetization that has only a longitudinal component and no transverse component, the effect of an RF pulse of angle other than 90 or 180 degrees (or multiples thereof) is to tilt the magnetization so that both longitudinal and transverse components are present. As above,for a B1 field aligned along the rotating frame x' axis, with an initial magnetization in the thermal equilibrium state:(M x ' (t )M y ' (t)M z ' (t ))=( 0?M eq sin(? B10 t )M eq cos(? B10 t ) )After a rotation of a nonzero magnetization that is initially aligned with the main applied field, by an angle that is not a multiple of 180 degrees, there will be a measurable signal due to the component of the magnetization in the transverse plane. As well, if the rotation angle was less than 90 degrees, there will remain a component of the magnetization in the longitudinal direction which will take less time to recover to its thermal equilibrium value than after higher rotation angles, as less of the longitudinal magnetization is removed by the lower angle excitation than is by a 90 degree excitation.153If a relatively rapidly repeating series (in comparison to the T1 time) of low-angle excitation pulses are applied to a magnetization, a steady state in the longitudinal magnetization may be established where losses due to the RF pulses and recovery from relaxation between the RF pulses are balanced.The magnitude of the precession signal in the steady state will depend on various parameters, including the T1 time of the object, the excitation angle, and the time between successive excitations. This relationship for the signal S is derived below:S?sin? (1?e?TRT1 )(1?e?TRT1 cos ?)where TR is the time between successive excitations, ? is the excitation angle, and T1 is the characteristic time of the object for longitudinal relaxation. Plotting the dependence of signal on ? for different TR/T1 reveals an interesting pattern, as seen in figure 5.9.Figure 5.9: Relative magnitude of signal in steady state cyclical excitation of magnetization as a functionof excitation angle. Signal curve is plotted for various TR/T1 ratios, as indicated by legend.154For large TR/T1, there is sufficient time between excitations for the longitudinal magnetization to effectively return to its thermal equilibrium value, Meq. In this case, the magnitude of the signal depends on the amount of magnetization rotated into the transverse plane, which is a sine function of the excitation angle that peaks at 90 degrees. As the TR/T1 ratio decreases, the signal magnitude drops (for fixed Meq) for all excitation angles, but drops more quickly at angles nearer to 180 degrees,where the magnetization has the largest distance to recover in the limited time, TR. Closer to 0 (or 360) degrees, very little of the longitudinal magnetization is removed by the excitation pulse, so short TR are sufficient for the magnetization to recover to near its thermal equilibrium value.More specifically, when acquiring an MR signal with a steady state series of excitations, the angle ? that produces the largest signal (and transverse magnetization after excitation) for a given TR/T1 decreases as that TR/T1 decreases. This may be derived by considering the angle where the derivative of the signal dependence on that angle is 0:dd? sin ?(1?e?TRT1 )(1?e?TRT1 cos?)=(e?TRT1?1)(e?TRT1?cos?)(e?TRT1 cos??1)2=0?=arccos(e?TRT1 )This angle is referred to as the Ernst angle. If the approximate T1 of an object is known, and a particular TR is required, using the Ernst angle for excitation pulses will give the largest possible signal.Although the largest overall signal magnitude, among all the curves plotted in figure 5.9, is observed when using a 90 degree pulse, this requires waiting for the longitudinal magnetization to fully relax between excitations. Rapidly acquiring signals with lower-angle excitations produces lower signals for each excitations, but allows signals to be acquired more rapidly. Viewed as signal quality per timespent acquiring, steady state methods such as this are sometime more efficient than waiting for full relaxation.1555.3 Spatial LocalizationMagnetic resonance imaging systems generally lack means to record signal arising from a specific location in an object while excluding signal from the remainder. Reception systems with multiple independent coils are sometimes used with a limited and localized volume of sensitivity, but MR images are generally produced with orders of magnitude more voxels than separate reception coils can be practically used. In many cases, only a single reception coil is available, yet it is still essential to produce an image, rather than just measure the sum signal from a distributed object.The most commonly used method in MRI for spatial localization is the application of spatial gradients of the main applied magnetic field during excitation, during signal reception, and between these stages. Gradient fields are generated by specialized gradient coils in the MR system. Typically there are multiple independently controlled gradient coils, allowing spatial information to be encoded in three spatial dimensions. This allows multi-dimensional images to be produced, rather than just measuring an average of an entire object.5.3.1 Frequency / Phase EncodingAs noted above, after excitation into the transverse plane, spins will begin to dephase due to local variations in their experienced magnetic field. A similar effect may be employed to intentionally cause transverse magnetizations to precess at different rates, by applying gradient fields that alter the magnitude (but not direction) of the main applied field after excitation and before or while the MR signal is recorded. Magnetic field gradients applied during precession cause variation in the rate of precession across an object, leading to accumulation of phase dependent on position. Unlike the localized small phase variations that lead to T2* or T2 signal decay, gradient-induced phase is, continuous, precisely controlled, and usually is applied in a linearly-varying strength, spread over theportion of an object being imaged.Applying linearly varying phase across an object is useful because it provides a means to probe the spatial distribution of the origin of the MR signal. Without any applied phase variation, and neglecting T2 effects, all magnetizations in the excited volume of an object will begin precessing with156the same phase. The measured signal by the MR system will generally be just the (reception-sensitivity weighted) summed signals from of all in-phase magnetizations in the excited volume, which provides no information about differences in signal arising from magnetizations at different locations in the object. Exposing the object to a gradient field produces patterns of phase accumulation across an object, which are useful for measuring the spatial distribution of magnetization. Most commonly used are spatially linear gradients, which produce a pattern of phaseaccumulation at a constant rate with distance in any fixed direction across an object.The phase that accumulates at any point, r, will be the time integral of the gradient field contributionat that point, G(r), and the complex magnetization C(t) may be written:C ( r? , t )=S ( r? )e?i ? ?t '=t 1t 2G ( r? , t ' )dt 'where the real and imaginary parts of C(t) represent the X and Y components of the magnetization in the fixed frame of reference. Alternatively, after a spatially linearly-varying gradient field producesa pattern of phase variation across an object, that pattern may be represented by a plane wave vector k:C ( r? )=S ( r? )e?i r???kThe signal that is measured by the MR system is the combination of all these excited spins simultaneously:C ( k? )=??rS ( r? ' )e?i r? '??k d r? 'This relationship is essentially the same as a Fourier transform in 2 dimensions, where the set of wave vectors k contains the spatial frequencies of the magnetization distribution being measured, and is referred to as "k-space". The signal is measured at a prescribed point in k-space by applying the appropriate gradients and measuring the signal produced by precessing spins in the object. By measuring a full set of k-space samples, the values of the Fourier transform of the object's spin distribution can be measured. To produce an image of the object, the inverse Fourier transform of the measured data is used.157A way to envision this process is to note that applying a varying phase, covering at least 2? range of phase spread, to a group of spins will lead to cancellation of the signal from those spins, unless the spins were already arranged with suitable distribution to cancel out the applied phase. Accordingly, applying a varying phase across an object will result in a measured signal proportional to the amountof spins initially in a distribution of phase with the same variation. If a varying phase is applied to a uniform distribution of phases - all aligned - the result will be no signal because the spin phases will be spread out. If a linearly varying phase is applied to a linearly varying phase distribution with a different spatial frequency, the spins will be further dephased and still produce no signal. But, if the linear variation of the phases initially in the object are matched and cancelled by the phase applied by a gradient, the result is a set of spins that are brought into the same phase by, and a relatively large measured signal. By this means, applying a linear gradient along a particular wave vector k results in signal proportional to the amount of that wave vector present in the initial spin distribution.5.3.2 Slice SelectionThe radiofrequency electromagnetic pulses that are used to excite spin systems to produce MR signals must oscillate at the Larmor frequency of the spins being excited. If a sufficiently off-resonance RF pulse is used, there is no resonance and spins are not excited. Any limited-duration RF pulse will have a finite spectral bandwidth, which limits the range of Larmor frequencies at which the pulse is able to excite magnetization.A spatially-varying magnetic field will produce a spatial variation in the Larmor frequency of spins. When a gradient field is applied, and the frequency bandwidth of an RF pulse is sufficiently narrow, alimited volume of an object in the scanner will have a matched resonance frequency to the applied RF pulse. Only within this volume will a pulse be resonant and able to rotate magnetization from the longitudinal direction into the transverse plane in order to generate measurable signal.This effect can be used to control the spatial distribution of excited magnetization; most commonly, this is used to limit the volume of an object from which signal is generated to a thin slice. The slice inwhich the RF field is resonant will be oriented perpendicularly to the gradient field direction, and 158will be centred where the main field and gradient field combined have a total field strength that produces a Larmor frequency that matches the RF field central frequency. By this method, a limited-extent slice of magnetizations may be excited, while the surrounding material is not. Alternatively, the spectral content of an RF pulse may be manipulated to produce more complicated slice profiles during excitation.5.4 T1 MappingT1-weighted imaging methods including inversion recovery or repeated application of low-angle excitation pulses and steady state were discussed above. These produces images whose contrast is strongly weighted by the T1 relaxation time of the local tissue. It is often useful, however, to quantitatively measure the T1 relaxation time, as it can be related to various properties of the tissue.T1 maps also describe only the T1 parameter, and will ideally eliminate all other sources of image contrast.Various acquisition methods for characterizing T1 in objects are discussed below.5.4.1 Inversion RecoveryInversion recovery may be employed to measure tissue T1 by varying the delay time between the initial inversion and the subsequent 90 degree excitation pulse. It is expected that the longitudinal magnetization will recover after inversion according to the relationship:M z(TI )=M eq?2M eq e?TIT1where Mz(TI) is the longitudinal magnetization at time TI after the inversion, Meq is the longitudinal magnetization after it has fully relaxed to its thermal equilibrium value, and T1 is the characteristic relaxation time for the longitudinal magnetization. Immediately after an (ideal) 180 degree inversion, the longitudinal magnetization will be the negative of its relaxed value; it has been inverted. The magnetization then relaxes towards its equilibrium value.159At any "inversion time", TI, after the excitation, a 90 degree excitation may be applied, and the resulting signal magnitude measured. The signal will be proportional to the amount of longitudinal magnetization immediately before the excitation. Varying the TI and measuring the dependence of the signal on it allows the above relationship to be used to extract the T1 time.Usefully, this method is somewhat insensitive to small variations or inaccuracies in the excitation andinversion angles, as long as these variations are consistent between samplings. If the inversion is slightly more or less than 180 degrees, the magnitude longitudinal magnetization after inversion may be less than before. The shape of the relaxation after the inversion will still have the same time dependence, although an extra parameter may be needed instead of using 2Meq .A problem with this method is that it can be extremely slow to acquire sufficient samples of the TI dependence of the signal in order to be able to reliably fit to determine T1. After each inversion and excitation, a sufficiently long time must be waited before the next inversion and excitation for the longitudinal magnetization to have recovered to approximately its thermal equilibrium value, Meq. A customary delay for this purpose is to wait 5 times the T1 relaxation time of the object. 5.4.2 Variable Flip-Angle Steady StateAs discussed above, a rapid low-angle excitation imaging sequence typically involves relatively rapid and low-angle excitation pulses being applied to an object. If the repetition time TR is fast compared with the T1 relaxation time, the amount of longitudinal magnetization present before excitation, andthus the signal strength after excitation, will tend to establish a steady state between recovery between excitations and loss to the excitations tilting spin away from the main field direction. This leads to a T1-weighted image, where faster longitudinal relaxation spins will have larger signal in the steady state.The dependence between relaxation time, repetition time, excitation angle and the magnetization (and thus signal strength) in a steady state can be derived. If M(TR) is the steady-state longitudinal magnetization after a recovery time TR since the last excitation, Meq is the fully relaxed magnetization, and M(0) is the remaining magnetization after an excitation of angle ?, then:160Excitation reduces the longitudinal magnetization:M (0)=M (TR)cos?Relaxation leads to recovery of the magnetization:M (TR)=M eq?(M eq ? M (0))e?TRT1M (TR)=M eq(1?E )+M (0)Ewhere E=e?TRT1Substituting for M(0):M (TR)=M eq(1?E )+M (TR)E cos ?M (TR)(1?E cos ?)=M eq(1?E )M (TR)M eq =(1?E )(1?E cos?)To convert from magnetization in the longitudinal direction to magnetization in the transverse plane after excitation, a factor of sin(?) is added to the expression for M(TR), the longitudinal magnetization before excitation:M xy=M eq sin ? (1?E)(1?E cos ?)And finally, the measured MR signal strength is proportional to this transverse magnetization:S?M eq sin? (1?E)(1?E cos ?)With this relationship, changes in ? or E (via changes in TR) can be used to change the signal from anobject, and different objects will have different E (via changes in T1). This provides a means to measure T1 values in an object, rather than just an image that has intensity weighted by T1. By 161acquiring images with multiple angles ("variable flip angle") or TR, the resulting signal curve can be fit with the known parameters to determine the object's T1.This method has limitations. The above discussion does not consider the fate of magnetization in thetransverse plane during the relaxation period between excitations. Because the sequence is rapid, there may be non-negligible transverse magnetization left by the time of a subsequent excitation. This magnetization will be manipulated by the following excitation, tilting some back into the longitudinal direction, and affecting the rate of recovery of that magnetization. This leads to an additional contribution to the signal in subsequent excitations, dependent on the T2* relaxation time of the object. This contribution will invalidate the relationship between signal, flip angle, TR and T1 discussed above. To avoid this, gradient spoiling may be used after signals are measured, to completely dephase the transverse magnetization, so that no net contribution to the longitudinal magnetization will occur in subsequent excitations.Additionally, it is not always possible to reliably and consistently control the angle of excitation with an MR system. RF excitation power levels (or the B1 field) vary across the volume of a scanner, due tocoil geometry and localized magnetic property variations (such as material susceptibility and discontinuities between tissues). As a result, a nominal flip angle will not produce the prescribed actual flip angle in all parts of an object. There are methods to measure the variation in B1 field strength across an object. Alternatively, multiple excitations with flip angles that differ by a known ratios may be used, and the relative B1 strength may be extracted from the resulting images. A scaling factor parameter may also be added to the steady-state short TR acquisition signal equation above, adjusting the actual flip angle at each voxel from the nominal angle for the whole object. In this case, the extra parameter must be fit along with the T1. These methods all have limitations and issues, and may not perform adequately for some accurate T1 mapping applications.1625.4.3 Look-LockerThe T1 mapping method used primarily in this work was Look-Locker readout during inversion recovery, with a flip-angle independent analysis method proposed by Chaung (2006).Look-Locker readout is a variation of the steady-state short-TR acquisition method discussed above, in which a series of low angle excitation pulses are applied to an object and the resulting signal tendstoward a steady state that is dependent on the sequence parameters and T1 relaxation rate. Unlike the steady-state method, in Look-Locker readout the signal is measured before it has reached a steady state, and the changing signal's dependence on number of applied excitations is (also) used to determine the T1. This provides a more reliable and faster measure of T1 dependence than does ashort-TR sequence read out after reaching a steady state.The Look-Locker readout scheme may also be combined with an initial 180 degree inversion pulse, similar to that used in a traditional inversion recovery sequence. The sequence begins with a non-selective inversion pulse across the whole object, which is followed by a short-TR slice-selective series of small angle excitations. As in standard inversion recovery, the resulting signals depend on the recovering longitudinal magnetization, but that signal is now read with multiple samples per inversion.Unlike a standard inversion recovery, Look-Locker's short TR readout also modifies the recovering magnetization, leading to a steady-state between sampling loss and relaxation after transient time dependence decays. Accordingly, the shape of the recovery curve is different than that of a single-measurement inversion recovery scan, and Look-Locker-specific data analysis is required.5.4.3.1 Look-Locker Signal DependenceWith Look-Locker inversion recovery, the magnetization and signal change with time depends on a combination of T1 recovery and the effects of the sampling pulses, leading to an apparent relaxationtime T1* after the inversion pulse that combines these effects.163Sampling pulses have the effect of multiplying the longitudinal magnetization by a factor dependent on the sampling pulse rotation angle ?. Magnetization before M(n) and after M(n)' the sampling pulse are related by:M (n)'=M (n)cos?letting y = 1 - cos ?:M (n)'=M (n)(1? y)After the sampling pulse, the magnetization relaxes at a rate determined by its T1 time, for a duration ? that is equal to the time between sampling pulses. If Meq is the fully relaxed longitudinal magnetization:M (n+1)=M eq?(M eq?M (n)' )e? ?T1=M (eq)(1?e??T1)+M (n)' e??T1letting u = exp(-?/T1) and substituting for M(n)':M (n+1)=M eq(1?u)+M (n)' u=M eq(1?u)+M (n)u(1? y)By induction as described by Look and Locker (1970), it can be shown that this leads to the relationship:[M (n)?M (?)]=[M (0 )?M (?)]e?n ?T1 (cos?)nWhere M(?) is the pseudo-steady state magnetization that occurs when magnetization loss due to the application of one sampling pulse is equal to magnetization gained during the sampling period ? due to T1 relaxation. This may be rewritten:M (n)?M (?)M (0)?M (?)=e?n ?T1 en ln(cos?)164=e?n?T1+n ln(cos ?)=e?n ? (1T1?ln (cos ?)? )This form illustrates that the decay of the magnetization, measured at times immediately before the magnetization is sampled, has an effective decay constant T1*:1T1?=1T1?ln (cos ?)?M (n)?M (?)M (0)?M (?)=e? n ?T1?This relationship may also be rewritten to more closely resemble the unaltered relaxation of an excited magnetization, with the magnetization M(n) immediately before each sampling pulse, the number n of pulses that have been applied, the magnetization immediately before a sampling in the pseudo-steady state that appears after many pulses, M(?), the magnetization before the first pulse M(0), and ? and T1* as above:M (n)=M (?)?(M (?)?M (0))e?n ?T1?This relationship may be fit directly to the measured signal intensity of a Look-Locker inversion recovery experiment to determine an object's T1*, with M(0) and M(?) as additional free parameters. The T1* may then be converted to a T1 using the known flip angle ?. In practice, however, the flip angle is not precisely known in objects, and may vary within an object, due to B1 field variation. B1 mapping measurements similar to those discussed above during scanning may be used, but may introduce inaccuracies in the resulting T1s.1655.4.3.2 Flip-Angle IndependenceThe method proposed by Chaung (2006) and used in this work eliminates the need for accurate knowledge of flip angle in a Look-Locker inversion recovery experiment. The method takes advantage of a relationship between M(0) and M(?) that is derived from the cycle of inversion, readout sampling, and delay before the inversion of the subsequent repetition of the whole sequence. Letting n = ? after many excitation and relaxation iterations:M (?)=M (n+1)=M (n)M (?)=M eq(1?u)+M (?)u(1?y )M (?)(1?u(1? y))=M eq(1?u)M (?)=M (eq) (1?u)(1?u(1? y ))=M eq (1?e? ?T1 )(1?e??T1 cos?)This may be further simplified using the relationship between T1, T1*, and cos ?:1T1?=1T1?ln (cos ?)?e??T1?=e??T1 cos?Which leads to:M (?)=M eq (1?e? ?T1 )(1?e??T1? )(Equation 1)166Additionally, a single iteration of the inversion recovery Look-Locker sequence consists of an initial 180 degree inversion pulse, a delay of time Td, a series of N sampling pulses of angle ? that are separated by a period ?, and a final delay Tp before the subsequent inversion. The total time TR for a single repetition of this sequence is:TR=Td +(N?1) ?+TpThe longitudinal magnetization has value M(TR) at the end of this sequence where M(N-1) is the magnetization immediately before the Nth sampling pulse.M (TR)=M eq(1?e?TpT1)+M (N?1)(cos?)e?TpT1 (Equation 2)M(TR) is typically substantially less than Meq due to relatively short time allowed for relaxation after the sampling pulses before the next inversion.After inversion, the magnetization initially has value -M(TR), and then relaxes for time Td until it reaches its value M(0) immediately before the first sampling pulse:M (0)=M eq(1?e?TdT1 )?M (TR)e?TdT1 (Equation 3)Substituting equation 2 into equation 3:M (0)=M eq(1?e?TdT1 )?M eq(1?e?TpT1 )e?TdT1?M (N?1)(cos ?)e?TpT1 e?TdT1=M eq(1?2e?TdT1+e?Tp+TdT1 )?M (N?1)(cos?)e?Tp+TdT1Substituting equation 1 for Meq:M (0)=M (?)(1?2 e?TdT1+e?Tp+TdT1 )(1?e?T1?)/(1?e?T1 )?M (N?1)(cos ?)e?Tp+TdT1(Equation 4)167As above, cos ? is related to the T1* and T1 times:e??T1?=e??T1 cos?cos?=e? ?T1?e??T1As noted by Chaung (2006), the relationship in equation 4 contains only known sequence parameters (?, Td, Tp, N) and parameters that may be determined by fitting the signal recovery curve(T1*, M(?), M(0)), and the unknown T1. After fitting for the parameters, the relationship may be used to calculate T1, without need for an accurate sampling flip angle ?. This method is used for fitting in T1 mapping in the experiments in this work to assess accumulation of Mn in brain tissue, asit was found to be the most reliable and fast method of those tested.1686 Manganese-Enhanced Magnetic Resonance ImagingManganese-enhanced MRI (MEMRI) uses paramagnetic Mn2+ as a contrast agent to modify MR signal. This is useful because it increases image contrast and reveals structures that are otherwise not visible, or are not visible with a particular imaging method. Applications are primarily in brain imaging in small animals, and may be localized to a particular structure such as a nerve fibre tract or active brain region, or may involve the whole brain.This section describes the mechanisms for Mn uptake and signal enhancement, published applications for MEMRI, various complications that can arise when using MEMRI, and experiments conducted as part of this work to examine dependence between MR image parameters and Mn concentration.6.1 Manganese UptakeMany MEMRI studies use Mn2+ for its ability to act as an analog for calcium in neuronal tissue, due toit having a similar ionic radius (Lin, 1997). Voltage-gated Ca channels take up available Mn2+ when active, causing the Mn2+ to be transported into interstitial extracellular space from the blood stream or cerebral spinal fluid (CSF) (Kuo, 2005) and thus accumulate in active neuronal tissue. In tissue, Mnlocally modifies MR image contrast acts (see below), which allows Mn accumulation in vivo and its resulting appearance in MR images to act as a marker for local neuronal activity.Notably, MEMRI is not sensitive to vascular signals, but rather is sensitive to "events directly related to cellular depolarization" (Tambalo, 2009). This is a notable advantage over, or distinction from, other activity-related MR contrasts such as the blood oxygen level dependent (BOLD) contrast or arterial spin labelling (ASL) perfusion imaging, which are at best indirect measures of neuronal activity, through its modification of local blood flow.Also notable is that Mn transport into the brain from the blood is obstructed and slowed by the blood-brain barrier (BBB) (Aschner, 1993; Takeda, 2003). The BBB is a biological system that impedes169the exchange of atoms and molecules between the CSF and blood. The barrier includes specialized endothelial cells that form a continuous layer with tight junctions, unlike the endothelial cells in most other tissues. This layer prevents or obstructs passing of substances between the blood and brain (Aschner, 1991), and plays an essential role in maintaining brain health (T?trault, 2008).After systematic administration by intraperitoneal or intravenous injection, most Mn is rapidly filtered from the blood plasma by the liver and secreted into bile (Aschner, 1991). Relatively little systemic Mn is transferred into the central nervous system (Aschner, 1993) and becomes available in brain tissue (other than the pituitary gland, which is outside the BBB). This is problematic because BBB crossing is relatively slow; enhancement is observed to peak several days after injections, making the technique more difficult to use for short-term brain activation experiments.Mn2+ uptake will be somewhat different if chelated, as discussed below, with a variety of uptake or biodistribution mechanisms, similar to the more commonly-used MRI paramagnetic contrast agents based on chelated Gd3+, such as gadobutrol.6.2 MEMRI ApplicationsManganese has been used as a contrast agent for numerous MRI studies. A brief review of its established uses follows.Published applications for activity-related Mn2+ update include localized uptake in the rodent brain due to electrical forepaw stimulation or administration of cocaine under anesthesia (Lu, 2007). In another experiment, animals were exposed to aural stimulation in the form of various frequencies ofsound, before or after administration of Mn to an animal, which lead to distinctive patterns of signal enhancement in images attributed to Mn accumulation (Yu, 2005; Yu, 2008). In another, the effect oflight or dark adaptation on the uptake of Mn in the rat retina was examined (Berkowitz, 2006). In another, MnCl2 was injected into the nostrils of rats, after which it distributed to various brain structures (Gianutsos, 1997).170A somewhat distinct application of Mn imaging is neuronal tract tracing (Silva, 2004; Pautler, 1998; Murayama, 2006). Mn is transported along neuronal tracts in vivo, and very localized progressive signal enhancement can be seen in MR images. In one case, Mn was injected into the eyes of rodents, and the optic nerve tract was observed to be enhanced in MR images (Pautler, 1998). A related technique for assessing brain connectivity (Pelled, 2007) involved injecting MnCl2 on one sideof the brain and assessing signal change on the opposite side in a Parkinson's disease rat model.Systemically administered Mn also acts as a non-specific contrast enhancer in T1-weighted MR images, as seen in this work in the MR Imaging section (see page 186). Systemic MnCl2 administration has also been used for brain structural characterization (Angenstein, 2007).Published MEMRI experiments generally report results in one of two ways. First, signal enhancementis observed on T1-weighted images such as a nerve tract tracing (Silva, 2004; Pautler, 1998; Murayama, 2006) or nonspecific contrast enhancement throughout the brain (Aoki, 2004; Angenstein, 2006). Secondly, P-value maps (Soria, 2008; Lu, 2007) are generated of brain cross sections that indicate areas where experimental data are highly consistent with there being an accumulation of Mn. In some cases (Kuo, 2005), region of interest-averaged T1 values or calculated Mn concentrations (Tambalo, 2009) have been reported in brain structures, but without sub-structural or pixel-level concentration estimates.MR Imaging of Mn2+ and Mn3+ with T2 measurement has been reported used for chemical reaction monitoring (Britton, 2006).6.3 Paramagnetic RelaxationThe magnetic moment of an electron, ?e = 5.79x10-5 eV/T, is 658 times that of a proton,?p = 8.80x10-8 eV/T. Interactions with electrons spins can produce strong coupling and relaxation of proton spins. However, like spins of protons and neutrons in atomic nuclei, electrons spins tend to form pairs of opposite orientations in an applied magnetic field due to the Pauli exclusion principle, and produce no net magnetic moment. As such, MR proton-spin electron-field relaxation effects can only occur with unpaired electrons.171Most stable molecules have no unpaired electrons, and thus produce no substantial electron-induced relaxation of proton spins. In biological tissues, exceptions are free radicals (Mi?ville, 2010), including metallic ions such as Gd3+ with 7 unpaired valence electrons, and notably for this work, Mn2+, with 5 unpaired valence electrons. Interestingly, free radicals can also be detected with electron spin resonance (ESR) (Hiramatsu, 1995), which is analogous to nuclear magnetic resonance (NMR), but detects transitions between energy levels of electron spin alignment.In aqueous solution, such as most biological tissue, polar water molecules tend to align in concentrichydration layers around ions such as Mn2+, with the water negative poles preferentially facing the positive ion. Proton Spin-lattice (T1) and spin-spin (T2) relaxation is increased primarily due to dipole-dipole interactions between the protons spins of these water molecules as they interact with the fluctuating magnetic field of the unpaired electrons. This field falls rapidly with distance, so that protons in water in close proximity to the ion are most strongly affected, although protons diffusing through the surrounding bulk solution are affected as well. Water molecules also move between hydration layers and the surrounding solution, allowing additional water molecules to be closely exposed to the ion field and undergo relaxation. The detailed description of these interactions is too complicated to fully discuss here, but includes other factors such as the rate of exchange of water or protons of water molecules between bulk and hydration layers, the proton and electron Larmor frequencies, rotation of molecules, number of molecules in hydration layers, and others (Caravan, 2009; Gore, 2013; Merbach, 2013; Pesaresi, 2010).Changes in R1 and R2 relaxivities (the inverse of the T1 and T2 times) due to ions such as Mn2+ or Gd3+ in dilute solution are generally proportional to the concentration of those ions (Pesaresi, 2010; Merbach, 2013). As such, T1-weighted MR images with short repetition times will generally show higher signal where these ions are present at suitable concentrations. Because the R2 relaxation rateis generally faster than the R1 rate in MR studies of biological tissues, the increase in R1 is generally the more-dominant effect, and Mn and Gd are treated primarily as signal enhancers. However, at very high concentrations of ions (including in some images acquired as part of this work), T2 or T2* effects may become important, which can lead to reduction in MRI signal or necessitate use of extremely short echo times (Nofiele, 2013).  1726.4 MEMRI ComplicationsSeveral complications affect the implementation of Mn as an MR contrast agent, particularly in the brain. Two of these, the blood brain barrier and Mn toxicity are discussed in this section.6.4.1 Blood Brain BarrierAs noted above, the blood brain barrier (BBB) obstructs transport of systemically administered Mn into brain tissue from the blood. This limits the usefulness of systemically administered Mn for short term activation studies or general brain contrast enhancement.In order to get more Mn into brain tissue more rapidly, BBB-breaking methods have been employed in animal models. These methods temporarily disrupt the BBB while Mn is administered, allowing much more Mn to rapidly enter brain tissue. Reported BBB-breaking methods include mannitol and ultrasound. Mannitol injections are typically done by surgically placing an injection port in the carotid artery and administering mannitol directly into the brain blood stream (Lin, 1997). This can result in a fairly widespread unilateral disruption of the BBB.Alternatively, ultrasound uses an external transducer to supply ultrasound energy to the brain, and injection of micro-bubbles into the bloodstream (Howles, 2010; Yang, 2010). The ultrasound pulse can be focused relatively tightly in the brain, where it interacts with the bubbles to disrupt the local BBB.Alternatively, the amount of Mn that reaches the brain may be increased by giving larger doses of Mn systemically. However, this is problematic because systemic injection of Mn in sufficient amountsto give useful enhancement throughout the brain can have severe toxic effects in animals (as discussed below).1736.4.2 Manganese ToxicityIt has been long known that manganese has potential negative health effects (Aschner, 1991). Chronic exposure to Mn occurs in trades such as mining and welding, and long-term exposure to Mn through the lungs in these jobs has been reported to lead to a Parkinsons-disease-like condition known as manganism (Zhang, 2003). Mn2+ accumulation in vivo occurs preferentially in mitochondriain areas associated with symptoms of manganism, and has numerous other potential interactions with cellular systems (Aschner, 1991). However, there is also evidence that the Mn3+ oxidation state is the cause of neurotoxicity and degeneration particularly in the striatum. As well, clearance of Mn2+by excretion in bile is rapid, while Mn3+ is eliminated more slowly (Aschner, 1991). Additionally, Mn may act as a calcium channel blocker, affecting physiology of muscles, and may have mutagenic or teratogenic (fetal) effects even when initially chelated (Bertin, 2009).Accordingly, it is generally not possible to conducted MEMRI experiments with unchelated Mn in humans. In rats, single subcutaneous MnCl2 injections of 0.5 mmol/kg (80 mg/kg MnCl24H2O) are reported to cause mild acute neurophysiological and behavioural changes, and may compete with Ca2+ for voltage-gated Ca channels in the brain (Eschenko, 2010). Single IP injections of 100 mg/kg lead to signs of cerebral toxicity, including reduced weight gain, brain volume decrease, and anxiety and depressive behaviours (Bouilleret, 2011). Single IP injections of 180 mg/kg MnCl24H2O in rats have been reported to lead to consistent tail necrosis, lethargy and poor grooming, while 2 fractionated doses of 90 mg/kg lead to occasional tail necrosis (Bock, 2008), while more fractionated IV doses consistently led to weight loss.Even in smaller amounts, Mn injected too rapidly may lead to acute animal death as seen in this work, likely because high Mn concentration affects the heart. Other procedural details may affect tolerance of the injections, including the carrier solution and its temperature, making conducting such experiments difficult. 174Chelated Mn has been used in humans, however, similar to chelated Gd contrast agents. Reported Mn-based agents include MnTTPS4 for liver tumour detection (Klein, 2005), MnDPDP as a hepatobiliary agent (Elizondo, 1991; Larsen, 1997; Tofts, 2010), and others (Bertin, 2009). The kinetics, particularly with regard to the blood-brain barrier, of these agents is likely quite different from those of unchelated MnCl2 / Mn2+, although the interactions of such contrast agents may be designed with a particular imaging purpose in mind (Bertin, 2009).6.5 Mn Concentration MappingAs noted above, published MEMRI studies rarely attempt to quantify the amount of Mn present in vivo. A major goal for this work was to introduce voxel-by-voxel estimates of Mn accumulation after administration. There is great potential for concentration-estimates to provide more information than is available from methods that only report the likely presence of Mn-related signal enhancement. Most importantly, concentration or distribution estimates can measure the relative strengths of enhancement, and thus be used to study the relative amount of neuronal activity in a brain region in response to stimuli.6.5.1 Relaxivity CalibrationTo quantify Mn concentration changes in vivo, this work observes changes in the R1 relaxation rate of tissue between baseline and after administration of Mn. Theoretical treatment (Caravan, 2009) and in vitro measurements (K?yl?, 2009; Yilmaz, 1999) suggest a linear relationship between change in Mn concentration and change in R1 relaxation rate of solutions. Testing the linearity of this relationship in vivo is an important aspect of this work. In particular, it is useful to examine whether the same relaxivity relationship applies across an image, or in which conditions the linearity breaks down.Additionally, if the Mn concentration change to R1 change dependence is sufficiently linear, it shouldbe possible to use a calibration factor to convert the measured R1 change to an estimated Mn concentration change. This would facilitate comparisons between separate experiments. This calibration factor is not known a priori, and is expected to be dependent on numerous factors. 175Various publications (see below) have reported Mn relaxivity concentration dependence results in varied conditions, with a variety of results. As such, in order to determine a calibration factor to use in this work, on the UBC Bruker 7T MRI system, it was necessary to measure an approximate calibration factor based on in vitro measurements.These calibration data were acquired by preparing vials of Mn solution in saline. 4.17 mM (0.53 mg/ml MnCl2) solution was prepared in lactated ringer's saline, and serially diluted by halves to produce a series of MnCl2 solutions down to 33 ?M (4.1 ?g/ml). Saline with no added Mn was also prepared. These solutions were placed in glass vials, which were placed in a solenoid coil near the bore centre of the UBC Bruker 7T small animal MRI system (which was the same scanner used for animal imaging; see page 181). A temperature probe was also placed in a similar vial near the same to measure temperature during data acquisition.The T1 time of the solutions in the vials was measured using non-imaging inversion recovery (see page 158). Different TI delays are used in multiple acquisitions, providing data showing the relationship between inversion time and signal intensity for each sample. This relation is readily fit toan inversion recovery function, in which one parameter is the solution's T1 time. This is considered to be a gold-standard method for measuring the T1 time (or R1 rate) in uniform samples such as this.To measure the effect of solution temperature on the relationship between Mn concentration and R1 time, the samples were prepared at temperatures between 37.7 and 44.9 C, and allowed to cool in the scanner while inversion recovery data were being acquired. During the time it took to acquire a single set of inversion recovery data for a sample, the temperature dropped by amounts between 0.5 C and 3.1 C.The largest drops in temperature during a single acquisition occurred when samples were at a highertemperature, when the rate of heat loss to room temperature air was greatest. As well, the duration of acquisitions and the time delay between starting successive acquisitions was adjusted to accommodate the varying relaxation rates in samples with different concentrations of Mn. As expected, samples with higher Mn concentrations had more rapid relaxation, and it is necessary to allow samples to fully relax to their thermal equilibrium magnetization prior to the start of each 176acquisition for consistent results. Due to the wide range of Mn concentrations used, fit T1 rates had an 85-fold variation across samples in this experiment. Acquisition repetition times were kept at least 5 times the T1 time of the sample, but were kept as short as possible within that limit to maximize the number of acquisitions and the rate at which they were acquired. The shortest time for a single acquisition is preferable because it minimizes the variation in temperature during the acquisition, ensuring that the multiple inversion-time delays are probing as consistent a relaxation curve as possible. More rapid acquisition is also useful to more thoroughly characterize the temperature dependence of the relaxation rate of the samples.Calibration data were plotted in figure 6.1 as R1 against MnCl2 concentration for samples at 37.7 C, a temperature similar to animal body temperatures, and at 31.5 C, somewhat lower than body temperature, and well outside the range at which data would be acquired in vivo. Figure 6.1: In vitro saline R1 dependence on MnCl2 concentration at two temperatures.177These plots formed straight lines (R2=1.000), with relationshipsAt 31.5 C:R1 = 4.974[MnCl2] + 0.186At 37.7 C:R1 = 4.204[MnCl2] + 0.228where [MnCl2] is concentration in units of mM and R1 in units of s-1.Published relaxivity concentration dependencies for MnCl2 include 3.59 s-1mM-1 in vitro water at 20 Cand 4.7 T (Bertin, 2009), 3.74 +/- 0.60 s-1mM-1 in vitro agarose gel, 5.15 +/- 0.78 s-1mM-1 ex vivo rat brain tissue at 37 C and 4.7 T (Tambalo, 2009), and 7.4 s-1mM-1 in vitro water at 3 T (Nofiele, 2013)These results demonstrate the importance of controlling temperature while acquiring R1 data. Changing temperature leads to changes in both the baseline Mn-independent relaxation rate of the solutions, as well as a change in the dependence of R1 on Mn concentration. Such temperature dependence of relaxivity is not surprising, as various published articles theoretically predict its effect. Measurement of its impact is poorly reported, however, in comparison with other factors thatimpact the relaxivity concentration dependence. To better illustrate the temperature effect, the measured R1 values for each Mn concentration were plotted against temperature, as seen in figure 6.2. Linear models fit to these data, and showed consistent linear dependence (R2 > 0.994) of R1 on temperature.178Figure 6.2: In vitro saline R1 dependence on temperature at various MnCl2 concentrations.For future in vivo experiments, calibration factors from these data were used to convert change in R1to change in Mn concentration.  The calibration factors were linearly interpolated between slopes at 31.5 C and 37.7 C to determine a factor for the temperature of the animals for each scan, as measured with temperature probes and as discussed in the Magnetic Resonance Imaging chapter. Linear interpolation in this manner was considered reasonable because the above figures show that relaxivity dependence in linear in both temperature and concentration of Mn.Notably, it can be seen in figure 6.2 that samples with higher Mn concentration and thus larger R1 were measured more frequently. As noted above, this was the case because samples with low R1 have correspondingly long T1 times, and required longer delays between samples to ensure full relaxation of the longitudinal magnetization between repetitions.1796.5.2 Other Calibration Factor LimitationsWhile in vitro experiments discussed above showed very linear dependence of R1 change on change in Mn concentration or temperature, it is not expected that these relationships will be fully maintained in vivo; numerous other factors may affect the dependence. As well, there are issues with the practicality of measuring R1 values in some cases, particularly when dealing with locally high concentrations of Mn in the post-injection images.A limitation of using relaxation rate change to measure Mn concentration is that MRI signal is most strong dependent on the presence of Mn in the 2+ oxidation state, but less so on the 3+ oxidation state. In blood plasma, Mn2+ may be oxidized into Mn3+ (Aschner, 1993), which is unlikely to be occurring in vitro saline. As well, it is specifically the Mn2+ state that mimics the effects of Ca2+ in its interaction with voltage gated calcium channels (Aschner, 1991; Lin, 1997). The T1 relaxivity of Mn3+ is rarely discussed in MEMRI publications, although its T2 relaxivity is approximately 8 times less than that of Mn2+ (Britton, 2006) suggesting a similarly small effect on the T1.Additionally, the presence of other solutes in solution can affect the relaxivity dependence of Mn. Binding of Mn2+ to macromolecules can strongly affect its relaxivity in vivo, with increases of up to one order of magnitude (Silva, 2004). Similarly, solutions containing proteins tend to have higher relaxation rate, and higher (but still linear) relaxivity dependence on Mn concentration than does water (K?yl?, 2009).Cr3+ can also contribute to relaxation rates in MRI, which may be a concern in this work due to possible contamination of the injected solutions with Cr. The T1 relaxivity of Cr3+ is reported to be an order of magnitude less than that of Mn2+ in serum (Yilmaz, 1999) or in solutions of several blood proteins including albumin (K?yl?, 2009).A rigorous discussion of potential Mn chemical and physical pathways after administration and their potential to impact interpretation of MEMRI results is beyond the scope of this work, but it remains a potential limitation. Notably, however, a major goal of this work is to compare Mn concentration maps derived from PET and MRI. The Mn-52 PET tracer is prepared in HCl, so is itself expected to be 180in the form of MnCl2 in the Mn2+ oxidation state. Accordingly, it may be reasonable to assume that similar pathways are present for Mn for both the MR contrast agent and the PET tracer, so the comparisons between modalities may remain useful despite potential uncertainty of the biochemistry involved or unreliability of the calibration factor between change in R1 and concentration of Mn in vivo.At high concentrations, Mn will substantially shorten T2* relaxation rate of tissue. Reported R2 relaxivity dependence on MnCl2 concentration include 48.42 s-1mM-1 in vitro water at 20 C at 4.7 T (Bertin, 2009), and 117 s-1mM-1 in vitro water at 3 T (Nofiele, 2013), which are substantially larger than the R1 relaxivity dependence as discussed above. R2 increase has the effect of reducing MR signal intensity in regions with high amounts of Mn. This effect is particularly strong when imaging with multiple signal measurements for each excitation (i.e. at longer echo times). With sufficient R2 increase, there may be insufficient signal after a sampling pulse to reliably measure the shape of the longitudinal relaxation curve during an inversion recovery Look-Locker experiment such as is used in this work. R2 increase may also sufficiently reduce signal so as to prevent acquisition of images usingmost standard anatomical MR imaging sequences. Ultra-short echo time (UTE) methods (Nofiele, 2013) may be used to extract more useful signal from objects with very rapid R2 decay, although thismethod may be difficult integrate into other imaging sequences. It may, however, be possible to use MR sequences that measure T2 instead of T1 at high Mn concentrations (Britton, 2006); like R1, the change in R2 is modelled as linear with concentration, so a similar technique via R2 change may be usable in this concentration regime in vivo.6.5.3 DiscussionThe calibration factors determined from in vitro experiments relating Mn concentration to relaxationrate change are used in the Magnetic Resonance Imaging and Multimodality Comparisons chapters to convert images of relaxation rate change to Mn concentrations. The impact of the limitations to the linearity between relaxation rate and concentration are investigated with comparisons between PET and AR images and MRI-derived relaxation rate changes in the Multimodality Comparisons chapter as well.1817 Magnetic Resonance ImagingMagnetic resonance imaging (MRI) experiments were done as part of this work in order to develop and test methods for measuring the distribution of manganese in live rats. As discussed in the manganese-enhanced MRI section of this thesis (see page 169), published uses of manganese-enhanced MRI (MEMRI) generally involve signal enhancement or statistical maps indicating probableaccumulation of Mn, and rarely report concentration changes. For this work, it was desired to estimate relative or absolute concentration changes and distributions of Mn with MRI, to potentially improve the experimental power of MEMRI studies, and to compare with other imaging modalities: positron emission tomography (PET) and autoradiography (AR).This section describes the UBC Bruker 7T MRI system, implementation details of the MR imaging methods used, and a description of the MR imaging experiments that were conducted, regarding their purposes and results. Appendix A contains a list of all animal imaging subjects, including those discussed in this chapter.7.1 7T MRI SystemThe MRI system used in this work is the UBC MRI Research Centre's Bruker Biospec 7.05 T small animal MR scanner. The MR system has a 30 cm internal diameter bore, which is reduced to 7 cm with gradient and radio frequency coil inserts in place. For proton imaging, the Larmour frequency inthis magnet is near 300.2 MHz. The magnet is superconducting and helium cooled. The gradient coils used in this work have a maximum strength of 400 mT/m.Images of rats were generally acquired using a volume coil transmit with actively-decoupled rat headsurface coil configuration, except for abdominal scans and select other scans that used a volume coil for transmit and receive configuration. Rats were mounted in a custom-built plastic bed, with teeth holder and in-built nose cone for anesthesia and air delivery, and were also held in place with adhesive tape during scanning. A rat being prepared for an MRI scanning session is shown in figure 7.1. Rats were anesthetized with 4% isoflurane in air prior to scanning, and maintained under 182isoflurane anesthesia as needed while in the magnet. Figure 7.1: Rat in MRI bed, with snout in anesthesia delivery cone (at right), surface coil tapedto head, signal transmission leads over its back, respiration monitoring probe attached toside, and rectal thermometer with lead in place (at left).Temperature was monitored using a rectal temperature probe (SA Instruments, New York, USA). Animal temperature was actively controlled by manually adjusting or computer control of the temperature of water passing through a heating pad surrounding the rat, or by adjusting the temperature of air being blown over the animal in the magnet bore. Temperature control was particularly carefully attended to during acquisition of temperature-sensitive relaxation rate R1 or T1maps.Respiration was monitored using a pressure sensor (SA Instruments) taped to the chest. The respiration monitor was also used for sequence triggering to minimize motion artifacts, and also to motivate adjustment of anesthesia concentration to keep animals unconscious but alive during scanning. 1837.2 R1 Relaxation Rate MappingR1 relaxation rate data were acquired using the flip-angle independent Look-Locker inversion recovery method developed by Chaung (2006, see page 162 for theoretical basis). Details of a typicalLook-Locker sequence include: volume-coil transmit, surface-coil receive, total sequence repetition time TR=10 s, echo time TE=3 ms, inter-excitation time=150 ms, 40 images per inversion, excitation angle ?=20 deg, initial inversion delay TI=18 ms, matrix size=128x72, field of view FOV=4.0x2.25 cm2,slice thickness=625 ?m, 17 slices, total acquisition time=12 min.Look-Locker image data were fit to generate R1 relaxation rate maps as described by Chaung (2006). Initial tests for IP MnCl2 injections were fit using software provided by Chaung. Beginning with the ICV MnCl2 injections discussed below, fitting for R1 was done using custom-written MatLab scripts. The use of custom scripts became necessary after the data acquisition software used on the Bruker 7T MRI system was reimplemented locally by 7T research scientist Andrew Yung, and after scanner hardware and software updates made Chaung's version unusable.Illustrations of the R1* fitting and R1* to R1 correction process are shown in figures 7.2 and 7.3 below. For each voxel, the series of Look-Locker signal measurements is fit with a 3 parameter modelthat provides an effective R1* value, which is then corrected to an R1 value by finding the solution toa nonlinear equation. The three parameter fit and the correction to determine R1 are discussed in the MRI Background Look-Locker Signal Dependence and Flip-Angle Independence sections (see pages 162 and 165).184Figure 7.2: Look-Locker R1* fitting to data for 3 parameter fit. MR single-pixel signal intensity (y-axis) plottedagainst inversion delay (ms, x-axis), with 3-parameter fit plotted as solid curve.Figure 7.3: Illustration of search for solution to non-linear equation used to correct R1* measured by MRI to R1, as discussed in theMagnetic Resonance Imaging Background Flip-Angle Independence section. Corrected R1 values are shown on the x-axis, while costfunction being optimized is shown on the y-axis. The corrected R1 value is found at the zero of the cost function. Additional anatomical images were acquired during the same scanning sessions using T1-weighted short-TR low excitation angle steady-state sequences or T2-weighted rapid acquisition with refocusing echoes (multiple spin-echoes per excitation) sequences.1857.3 Initial IP MnCl2 InjectionPrior to this work, the UBC 7T MRI group had little experience with using MnCl2 as an MRI contrast agent in rats, or with acquiring quantitative relaxation rate data in live animals. It was thus necessaryto acquire a series of MR images with systemic administration of MnCl2 to gain experience with the techniques involved. It was also considered important to validate that the expected contrast or relaxation rate changes from Mn administration could be detected and measured, and to provide data to compare with initial PET results.IP injections of MnCl2 are better tolerated by rats and are less likely to cause complications than are IV injections (Bock, 2008). IP injections can also be given with larger doses of MnCl2 and are less dependent on careful control of injection speed and solution temperature (Silva, 2004). Accordingly, IP injections were preferred for these tests.A rat (320 g) was anesthetized and had a needle placed in its abdomen to deliver MnCl2 solution IP, fed by a syringe and catheter. The rat was placed in the UBC 7T MRI scanner with this injection line attached, to allow injection between and during MR acquisitions, without repositioning the animal. The catheter allowed the injection to be driven by an experimenter without placing a hand directly into the magnet bore, which might have interfered with image acquisition. The MnCl2 solution was prepared by massing 0.633 g MnCl24H2O and mixing in 10 ml of phosphate buffered saline (PBS), giving a solution of 40 mg/ml MnCl2. After acquiring baseline T1- and T2-weighted images and T1 maps, 0.52 ml of MnCl2 solution was injected over 30 s. Additional T1-weighted images and T1 maps covering most of the brain were acquired over the next 2 hours, and again 21 hours, 2 weeks, and 3 weeks post-injection.Figure 7.4 shows a time-series of T1-weighted MR images of this animal, acquired before and after administration of MnCl2. As expected, no short term (20 minutes post-injection) change in contrast isseen in the brain, compared with the baseline image. By 100 min post-injection, localized signal enhancement is seen in  ventricles and the pituitary gland. By 1 day post-injection, the ventricular enhancement has faded, but contrast is enhanced throughout the brain. General contrast 186enhancement, compared with the baseline image, remains apparent, but decreases with time up to the 3 week post-injection image. These results are qualitatively consistent with those reported in literature (Aoki, 2004; Tambalo, 2009) after systemic administrations of MnCl2 in rats.Figure 7.4: Midsagittal slices of T1 weighted MR images of rat head. Images were acquired at baseline, andafter abdominal injection of MnCl2 at times indicated in the figure. In the 100 min image, ventricles (marked withwhite arrows) and pituitary (marked with yellow arrows) have pronounced localized signal enhancement. Pituitaryenhancement increases at 1 day, and remains visible in the 1 week image, and may be present at 3 weeks.Figure 7.5 shows a time-series of T1-weighted images acquired as part of a Look-Locker inversion recovery T1-mapping experiment. These images illustrate the raw data that is used as input to the T1fitting procedure discussed in the previous section (see page 183); for each pixel, the series of images acquired at different times after the inversion are combined into a time-varying signal, similar to that seen in figure 7.2. Notably, these images show the magnitude of the signal, while the signal curve in figure 7.2 has negative values for the initial time points, due to negating the signal values before the minimum magnitude point during pre-processing in order to reproduce the expected negative signal during those initial time points. As discussed in the MRI Background chapter, the initial signal samples are expected to be negative because the inversion recovery sequence initially inverts, or negates, the longitudinal magnetization, which leads to a 180 degree phase shift, or negation, of the transverse magnetization signal after sampling excitations. The pattern of decreasing signal, passing through zero signal at different times in different pixels (particularly in the 3rd and 4th shown time points), and subsequent recovery of signal with time after inversion is similar to images shown by Chaung (2006).  Baseline                                                     20 min                                                      100 min  1 day                                                          2 weeks                                                    3 weeks187Figure 7.5: Selected frames from sequence of T1-weighted images acquired as part of a Look-Locker inversionrecovery T1-mapping acquisition. Data are acquired in a multi-slice imaging sequence, and reconstructedto produce images for multiple delay times after the inversion, as indicated in the figure. This acquisitionwas done at baseline, before the animal received any MnCl2 injections.No short-term reaction to the MnCl2 injection was observed in this rat. However, at the time of the 21 hour post-injection scan, a lesion was observed at the site of the injection. Veterinarians were consulted, and it was concluded that this was likely a reaction to the injection, and that the injection - which was intended to be IP - may have be deposited in the subcutaneous space instead, which is more likely to lead to reactions. The rat was not otherwise observed to be in pain or sick, and it was considered by the veterinarian to not be necessary to euthanize the animal. Topical antibiotics were applied over the next two weeks, and the rat recovered well.        TI = 64 ms                      104 ms                          264 ms                         344 ms                        384 ms                                                624 ms                       1024 ms                       1224 ms                      1624 ms188The brain R1 maps at baseline and post injection times of 70 min, 120 min, and 1 day were manually coregistered (which required only translations in-plane, not rotations). Baseline R1 values were subtracted from post-injection images, and converted to concentration maps (see page 174). The resulting images are shown in figure 7.6.Figure 7.6: Midsagittal slice images of Mn concentration (scale in mM) in rat brain. Images are produced bysubtracting R1 relaxation rate data from two separate image acquisitions: one at baseline, and one post-injection.Maps are shown for the times indicated in the figure after IP injection of MnCl2. Accumulation was particularlyvisible in the pituitary gland, outside the blood-brain barrier (marked with white arrows).Short-term Mn accumulation was seen localized in ventricles within the brain, and in the pituitary gland. After 1 day, general increase in Mn concentration was observed throughout the brain. This is consistent with results seen in T1-weighted images.7.4 ICV MnCl2 InjectionsAn important result for this work is the comparison of MR methods for localizing Mn in the brain with other modalities (discussed above). However, the previously-discussed IP and IV MnCl2 systemicinjection experiments with rats had relatively little uptake of Mn in the brain (compared with the pituitary gland and the amount injected) making such comparisons difficult. As well, methods for blood-brain barrier disruption were not readily available for use during this work. Instead, it was decided to use the existing expertise in intra-cerebro-ventricular (ICV) injection that was available to bypass the blood-brain barrier (BBB) for Mn delivery. ICV injections are routinely given to rats as part 70 min                                                      120 min                                                     1 day189of other studies of brain function in rats, and the technique could be easily adapted for MnCl2 injections.ICV injections are a surgical procedure that is performed under isoflurane anesthesia, and are described in more detail in the Positron Emission Tomography Imaging ICV Mn-52 Injections section (see page 96).7.4.1 ICV PreparationsAs with systemic injections, it was necessary to research and test appropriate dosage and volume of injections for ICV delivery of MnCl2. Because Mn is impeded from entering the brain by the BBB, and the amount reaching the brain is orders of magnitude less than the amount given systemically, it is quite plausible that injecting similar amounts directly into the brain would have much more pronounced health effects. As well, the volume injected systemically - up to several ml - would be infeasible for injection directly into the brain. Some of these considerations also apply to ICV injections containing radioactive Mn-52, as discussed in the Positron Emission Tomography Imaging chapter.ICV injections with volumes up to 30 ?l were previously routinely given to rats with no reported health effects prior to this work as part of unrelated experiments. It was desirable to use the largest volume injection possible due to limitations on the concentration of Mn-52 solutions that could be prepared, and the need to have sufficient amount of Mn-52 for PET imaging purposes. Because the technique was already used routinely, it was decided that tests of saline ICV injections, without added MnCl2, were not necessary.Instead, MnCl2 solutions were prepared and injected ICV in rats to test this technique. A solution of MnCl2 was prepared by dissolving 991 mg of MnCl24H2O in approximately 30 ml of distilled, deionized water, of which 3 ml was then passed through a 0.2 ?m filter into a blister top vial. A second solution of MnCl2 was prepared similarly, by dissolving 990 mg of MnCl24H2O in 30 ml of phosphate-buffered saline (PBS), of which 3 ml was also filtered into a blister top vial. Lastly, an additional 3.5 ml of the second solution was poured into a beaker containing dry chromatography 190column eluent, which was prepared in a manner similar to Mn-52 injectable solutions as discussed inthe Tracer Production chapter, except without adding any Cr foil or Mn-52. The contents of the beaker were then similarly filtered into a blister top vial. This solution was intended to test for effects of injecting a solution containing any potential contaminants that may be present in the output from the chromatography column separation process that would then be injected with a Mn-52 preparation. 990 mg of MnCl24H2O at a molar mass of 197.91 mg/mmol gives 5.01 mmol MnCl2, which, when dissolved in 30 ml water, gives a solution of 167 mM MnCl2. A 5, 10, or 30 ?l injection would contain 0.834, 1.67, or 5.00 ?mol of MnCl2, respectively, which correspond to 105 ?g, 210 ?g, and 630 ?g MnCl2 (the maximum allowed by the protocol approved by the UBC Animal Care Committee (ACC)). Injections were given to rats of these solutions, as shown in table 7.1.Rat BodyMass (g)ICV injection167 mM MnCl2Reaction740 5 ?l ICV of MnCl2 in water none656 5 ?l ICV of MnCl2 in PBS none753 5 ?l ICV of MnCl2 in PBS reduced weight gain567 10 ?l ICV of MnCl2 and column residue in PBS reduced weight gain817 30 ?l of MnCl2 and column residue in PBS death while under anesthesiaTable 7.1 - Intra-cerebro-ventricular (ICV) injections of MnCl2 in water, phosphate-buffered saline (PBS),or PBS mixed with chromatography column eluent residue, into rats, and observations of reactions.From these results, it was concluded that the amount of Mn being injected, or the contents of the column residue, might contribute to negative health effects from ICV injections. In particular, ICV injections of  5 ?mol of MnCl2 are unsafe in rats, while 1.67 ?mol is relatively well tolerated.Future injections used 5 or 10 ?l injections of MnCl2 at concentrations similar to those tested here, or injections with lower concentrations and total amounts of MnCl2 at up to 30 ?l volumes. Because there was no alternative method available for production of Mn-52 injectable solutions, and it was not known to be problematic to use the available method, it was decided to proceed to imaging experiments with Mn-52 ICV injections despite potential for complications from column eluent contamination. 191Additional non-imaging experiments similar to those discussed in this section could have been conducted to assess any impact of such contaminants, however it was considered to be a more effective use of animal subjects to proceed to imaging experiments, in order to minimize the total number of animals used. As well, the animals used for the non-imaging experiments discussed in this section had no other planned experimental uses, and would have been euthanized without scientific benefit regardless of these tests. Conversely, additional testing to assess potential health effects of column contaminants would have required ordering additional animal subjects, and wouldhave not provided any direct benefit in the form of information relevant to the actual imaging goals of this study. As such, subsequent ICV injections were done with the knowledge that the amount injected might need to be reduced if acute reactions to the injections occurred. However, as discussed in the Positron Emission Tomography Imaging chapter, no such acute reaction after ICV injections including chromatography in the preparation of the solutions was observed. Regardless of these results, as discussed in the Tracer Production chapter, alternative Mn-52 production and purification methods should likely be explored for any future Mn-52 imaging experiments.7.4.2 ICV ImagingThree rats (305 g, 318 g, and 306 g) were anesthetized and scanned in the 7T MRI to provide baseline T1 maps and anatomical images. One day later, these rats received 5, 10, and 10 ?l ICV injections, respectively, of the same MnCl2 in PBS solution used for the ICV injection tests in the previous section, with a total MnCl2 content of 0.834 or 1.67 ?mol. One and two days post-injection, additional MRI images were acquired of these rats. All three rats had stable or slightly reduced weights over the subsequent two weeks, although it is notable that these rats also received multiple anesthesias for MRI scanning, unlike rats in the previous section which were not scanned. The purpose of these injections and scans was to verify that there was MR signal enhancement and measurably changed R1 in rat brains after ICV injections of this volume and amount of MnCl2. This was readily seen, as shown in figures 7.7 and 7.8. In the post-injection T1-weighted images, striking contrast and structural details are seen in brains, particularly in the cerebellum in sagittal slices, and various midbrain structures in the coronal slices, which are not seen in similar scans without contrastinjection. (While precontrast T1-weighted images were not acquired in this session, similar scans 192acquired previously and subsequently show markedly less structural detail.) This enhanced contrast is notably bilaterally symmetrical in the coronal oriented image slices. Figure 7.7:  Midsagittal slice of rat brain T1-weighted MRI acquired 1 day after ICV injection of MnCl2.Striking contrast enhancement is seen in the cerebellum (marked with arrow).Figure 7.8: Coronal slice of rat brain T1-weighted MRI acquired 2 days after ICV injection of MnCl2.Structural details are seen throughout the brain.T2-weighted images were also acquired at baseline and post-injection, as shown in figure 7.9. At baseline, the T2-weighted images already show more structural details than do similar T1-weighted images. Post-injection, some already-bright structures appear brighter, but there is not the dramatic change in contrast as seen with the T1-weighted images. Notably, in the T2-weighted image, there is unilateral darkening in the T2-weighted image on the side corresponding to the location of the ICV injection. This asymmetry is not seen in the T1-weighted images or pre-injection T2-weighted images. This darkening may indicate locally higher concentrations of Mn2+ are causing T2-related signal loss. That interpretation would be consistent with relative darkening seen in the colliculus of the T2-weighted images, which corresponds to a region that appears locally very bright in T1-weighted images, suggesting locally high accumulation of Mn2+. 193Figure 7.9: Coronal slices of (the same) rat brain T2-weighted MRI acquired at baseline (left) and 1 day after ICV injection of MnCl2.Unilateral darkening is seen near the site of injection (marked with arrow).R1 relaxation rate maps were also produced of the brains of these rats, at baseline and post-injection, as shown in figure 7.10. The relaxation rate seen in the post-injection images of these rats was relatively large: values up to 5 or 6 s-1 are seen in the brainstem and midbrain. These values were judged to be larger than ideal for this experiment, as an R1 of 5 s-1 corresponds to a T1 time of 1/5 s or 200 ms. This T1 time is problematically low because with a 17 slice Look-Locker acquisition, the sampling repetition time ? was limited to values around 150 ms or larger. With 150 ms sampling for a 200 ms T1, at most one or two samples would be taken before the signal had passed through zero, and it was judged that too few samples would be taken before the signal had effectively reached steady state. Accordingly, it was decided that using a lower concentration of MnCl2 would be preferable for future experiments, so as to have a lower R1 and more samples during the relaxation portion of the signal curve.194Figure 7.10: Midsagittal slices of (the same) rat brain R1 relaxation rate map. Scales in s-1. These maps weregenerated from data acquired before (top) and 1 day after (bottom) ICV injection of MnCl2.Strong increase in R1 relaxation rate is seen in red and yellow regions of midbrain and spinal cord.Moderate increase in R1 relaxation rate in cyan regions elsewhere in brain.7.5 Mixed Mn-52 and Nonradioactive Mn InjectionsTo facilitate comparisons of Mn imaging results between different imaging modalities (MRI, PET, and AR), animals were given injections that contained non-radioactive MnCl2 in amounts suitable for MEMRI, which also contained radioactive Mn-52 to act as a PET and AR tracer. These injections included both right-lateral ventricle targeted injections (ICV), and an IP injection, in separate animals. The results of the MR imaging on these animals is outlined below.1957.5.1 Mixed ICV InjectionsAs also discussed in the PET imaging section (see page 99), two rats (407 g and 428 g) received surgical ICV right-lateral targeted injections of Mn-52 mixed with 6.5 mM MnCl2 in 30 ?l phosphate-buffered saline. After imaging, it was noted that the injection was successfully placed in the right-lateral ventricle in only one animal. In the other, the injection missed the right-lateral ventricle, and was instead placed in brain tissue near that structure (see below).The injected volume was kept as large as possible for these injections - 30 ?l - in order to maximize the amount of Mn-52 injected, as concentrating it further was not possible. The concentration of non-radioactive MnCl2 was reduced compared with previous injections to allow this larger volume without causing animal death due to large mass of MnCl2 as occurred in initial tests. Specifically, 30 ?l volume with 6.5 mM MnCl2 contained 0.19 ?mol MnCl2, or approximately 12% of the amount given to rats in the previous section where problematically large relaxivities were observed.Before injections, these rats received baseline MRI scans to acquire anatomical images and to produce relaxation rate R1 maps of their heads. After injections, the rats received additional MRI scans, and were also scanned in the microPET. The rats were later euthanized by carbon dioxide and their brains extracted and imaged with autoradiography.For one of these rats, the T1 and T2-weighted post-injection images (shown in figures 7.11 and 7.12)generally resembled those of the ICV injected rats in the previous ICV MnCl2 imaging section (see page 191). Compared with baseline, localized signal enhancement is seen most prominently in the colliculus and other midbrain structures in T1-weighted post-injection images, although less dramatically so due to the lower amount of Mn2+ available. T2-weighted images look similar between baseline and post-injection. Both types of anatomical image are left-right symmetrical.196Figure 7.11: Coronal slices of T1-weighted images of rat that received successful ICV injection ofMnCl2 and Mn-52 into the right (bottom in image) lateral ventricle. Images were acquiredat baseline (left), 1 day post-injection (centre) and 4 days post-injection (right). Strong signalenhancement is seen in the colliculus (marked with arrow).Figure 7.12: Coronal slices of T2-weighted images of rat that received successful ICV injection ofMnCl2 and Mn-52 into the right (bottom in image) lateral ventricle. Images were acquiredat baseline (left), 1 day post-injection (centre) and 4 days post-injection (right).Interestingly, the injection route can be faintly seen in appropriate slices of both T1 and T2 weighted images, as seen in figures 7.13 and 7.14.197Figure 7.13: Coronal and sagittal slices of T1-weighted images of rat that received successful ICV injection ofMnCl2 and Mn-52 into the right lateral ventricle, acquired 1 day after the injection. These slicesare positioned to cut through the injection route (marked by arrows).Figure 7.14: Coronal and sagittal slices of T2-weighted images of rat that received successful ICV injection ofMnCl2 and Mn-52 into the right lateral ventricle, acquired 1 day after the injection. These slicesare positioned to cut through the injection route (marked by arrows).R1 maps for this rat were produced as well (shown in figure 7.15). The R1 distributions are left-right symmetric and are similar to those seen in previous ICV injections, with generally higher values post-injection compared with baseline, although in this case the peak values are closer to 1.5 s -1, compared with the large regions of values near 5 s-1 in the animals that received higher amounts of MnCl2 (e.g. as seen in figure 7.10). Given that baseline R1 values are near 1 s-1 in either case, and the changes in R1 as a result of these MnCl2 administrations are roughly 0.5 and 4 s-1. This is roughly consistent with the ratio of the amounts of MnCl2 administered as noted above, 12%, suggesting at least roughly proportional dependence of change in relaxation rate on tissue concentration of MnCl2 within this range. More quantitative comparisons of such changes in R1 relaxation rate are discussedin the Relaxation Rate Map Analysis section (see page 204).198Figure 7.15: Coronal slices of R1 relaxation rate maps of (the same) rat which received ICV injection of MnCl2 and Mn-52.Images were acquired at baseline and post-injection at times indicated in figure. Scale in s-1.For the other ICV mixed non-radioactive MnCl2 and Mn-52 injected rat, it is immediately apparent that the injection did not achieve the same result. It appears that the injection was not placed successfully in the right lateral ventricle. There is marked asymmetry in both T1 and T2 contrasts (shown in figures 7.16 and 7.17), and there is darkening along the injected-side anterior edge of the lateral ventricle in both T1 and T2 weighted images. In the T2 weighted only, there is brightening on the injected-side posterior edge of the ventricle and over the entire ventricle on the contralateral side. Slightly to the injected side of the space between the anterior portions of the lateral ventricles, there is a blotchy darkening in the T2 weighted post-injection image. Elsewhere in the T1 weighted images, there is slight symmetrical contrast enhancement in general midbrain structures similar to the other ICV injected rats.Baseline1 day4 days199Figure 7.16: Coronal slices of T1 weighted images at baseline (left), 1 day (centre), and 4 days (right)post-injection of rat that received misplaced right lateral ventricle ICV injection of MnCl2 and Mn-52.Reduced signal is seen in the area of the injected-side lateral ventricle (marked with arrows).Figure 7.17: Coronal slices of T2 weighted images at baseline (left), 1 day (centre), and 8 days (right)post-injection of rat that received misplaced right lateral ventricle ICV injection of MnCl2 and Mn-52.Reduced signal is seen in a region between the lateral ventricles (marked with arrows).The injection route is more prominent in the anatomical images for this rat as well, as seen in figures7.18 and 7.19.200Figure 7.18: Coronal and sagittal slices of T1-weighted images of rat that received misplaced ICV injection ofMnCl2 and Mn-52 targeted at the right lateral ventricle, acquired 1 day after the injection. These slicesare positioned to cut through the injection route, which is marked by the white arrows.Figure 7.19: Coronal and sagittal slices of T2-weighted images of rat that received misplaced ICV injection ofMnCl2 and Mn-52 targeted at the right lateral ventricle, acquired 1 day after the injection. These slicesare positioned to cut through the injection route, which is marked by the white arrows.R1 maps for this rat were produced as well, and are shown in figure 7.20. The ventricle contralateral to the injection site appears bright in the T2 weighted image, dark in dark in the T1 weighted image, and accordingly appears to have a reduced R1 compared with baseline. This is characteristic of edema, as cerebral spinal fluid (CSF) appears bright in T2 and dark in T1 weighted images. The dark region in between the lateral ventricles in the T2 weighted image corresponds to a region of very high R1, suggesting very localized MnCl2 deposition that was unable to spread throughout the brain through the cerebral spinal fluid in the ventricles as occurred for the rat with the successful ICV injection. The effect of edema on apparent R1 of tissue will be relevant later for comparison with other modalities.201Figure 7.20: Coronal slices of R1 relaxation rate maps of (the same) rat which received misplaced ICV injection of MnCl2 and Mn-52.Images were acquired at baseline and post-injection at times indicated in figure. Scale in s-1.Large increases in R1 are seen localized near the site of injection (marked with arrows).7.5.2 Mixed IP InjectionsAs previously discussed in the PET imaging section (see page 101), a rat (490 g) was given an IP injection containing Mn-52 and 75 mM MnCl2 in 2.05 ml PBS. This rat also received MRI and PET scans of the head at baseline (MRI) and post-injection (both), after which it was euthanized and imaged with autoradiography. MRI scans included T1 weighted, T2 weighted, and R1 maps in the brain. A large field of view body T1-weighted MRI scan was also acquired, covering from the cerebellum to the abdominal organs.Baseline1 day8 days202Baseline images of the brain are similar to other baseline scans discussed above. One day post injection, T1 weighted images (shown in figure 7.21) were similar to those of the rat that received anIP injection while in the MRI scanner at a similar time point discussed above (shown in figure 7.4). Slightly increased contrast is seen throughout the brain, with marked increase in the pituitary gland and a structure located inferior to the pituitary, exterior to the skull (at bottom of image field of view). Enhancement is much less pronounced and less widespread for the IP injections compared with the ICV injections, despite the much larger total amount of MnCl2, due to the uptake being slowed and inhibited by the blood brain barrier for the IP injections. T2 weighted images of this animal were unremarkable after injection.Figure 7.21: Midsagittal (left) and coronal (right) slices of T1 weighted images of rat brain 1day post IP injection of MnCl2 and Mn-52. Signal enhancement is seen in the pituitary in theT1 weighted image (marked with yellow arrow), similar to figure 7.4.The abdominal T1 weighted image (shown in figure 7.22) is difficult to interpret. The midsagittal sliceshows a long stretch of spinal cord extending from posterior to the cerebellum, curving down and back up as it heads in the posterior direction. There is a section in the middle of the image with pronounced artifacts arising from motion of the lungs and diaphragm during image acquisition, which interferes with spatial localization of spins. A large uniform organ is seen near the posterior end of the field of view, which appears to be the liver.203Figure 7.22: Midsagittal slice of abdominal T1 weighted image of rat 6 days after IP injection ofMnCl2 and Mn-52. The posterior brain (marked with white arrow) and spinal cord (marked withgrey arrows) are visible. Motion artifacts are seen in the centre of the image, likely due to breathing.The large uniform region (marked by yellow arrow) may be the liver.R1 maps were also acquired of the brain for this rat, as shown in figure 7.23. The differences between baseline and post-injection are notably less than seen for the ICV injected animals. A similar pattern is seen, however, with increased R1 in the middle of the cerebellum and various midbrain structures, but not the cortex. The olfactory lobes are notably brighter in these maps, which is not observed to proportionally the same degree in the ICV injected animals. The pituitary gland also has a large change in relaxation rate, consistent with other images showing signal enhancement or relaxation rate change, although is not seen in these coronal image slices. As well, in these images, R1 values are relatively stable over 5 days post-injection in this animal, whereas in the ICV injected animals, the R1 values tended to be higher in the early post-injection scans. This may indicate that at higher concentrations, the biological mechanisms that clear Mn from within thebrain are more active.204Figure 7.23: Coronal slice R1 maps of rat that received IP injection of MnCl2 and Mn-52 at baseline (upper left),1 day (upper right), 4 days (lower left), and 6 days (lower right) post IP injection. Scale is in s-1.More detailed analysis of changes in R1 shown in the relaxation rate map images shown in these figures are discussed below.7.6 Relaxation Rate Map AnalysisDirect comparison of Mn imaging results between MRI and other modalities requires converting separately-measured baseline and post-administration R1 maps into R1 difference maps. As noted earlier, it is the difference in the R1 relaxation rate that is expected to be roughly proportional to the concentration of Mn that has accumulated in tissue, and comparison with PET and AR will be used toexamine the degree to which this relationship holds.To create such difference maps, R1 maps from post-injection and baseline states must be coregistered so that anatomical features in each are located in the same voxels. The values of matching voxels in the baseline map are then subtracted from the post-injection map to generate a change in R1 map. As discussed above, the first concentration maps (shown in figure 7.6) produced from an IP injection of MnCl2 while a rat was in the MRI scanner were analyzed using only 2D translations for coregistration of baseline and post-injection T1 maps. For those images, it was not 205necessary to do any more complicated data processing to produce R1 difference maps, which were then converted to concentration maps using a scale factor. The images of rats that received ICV and IP injections of mixed non-radioactive MnCl2 and Mn-52, required a more complicated registration procedure, however, due to more complicated translations and rotations between the positions of the anatomy in the images.This section describes the coregistration process and issues that arose during it. Finally, the resulting R1 difference maps for rats that received mixed MnCl2 and Mn-52 injections are presented.7.6.1 CoregistrationBaseline and post-injection R1 maps for mixed-injection rats were interpolated by averaging between planes to double the number of planes of data and reduce the plane spacing from 0.625 mm to 0.3125 mm. 0.3125 mm is also the in-plane pixel size (40 mm / 128 voxels), giving an isotropicvoxel size after this resampling. Having an isotropic voxel size was useful for doing coregistrations, which involved arbitrary rotations and translations in 3D space.Coregistrations were done by importing the 3D R1 map data as images into the PET analysis software ASIPro (Siemens / Concorde), which is used for analysis for microPET images. This software has a convenient feature for entering manual translations and rotations while overlaying two images, which was useful for this coregistration task. The registration was optimized by visual inspection, without use of a numerical cost function. As discussed in the Autoradiography Slice Registration section (see page 125), manual registration was preferred due to the relatively small number of volumes being registered did not require an automated method.The coregistration process involved loading a moving and target image, determining a 3D transformation to match the moving images to the target image, and then applying this transformation to resample the moving image by trilinear interpolation. This resampling introduced a blurring effect on the moving image, as illustrated in figure 7.24, which was a potential problem given that the purpose of this resampling is to find the difference in R1 values between the moving and target images at corresponding locations. Blurring one of these images removes sharp details 206and contrast in the images, which will not be matched in the un-resampled target image, potentially leading to artifacts in the difference image.Figure 7.24: R1 map coronal slices before (left) and after (right) resamplingfor coregistration. Loss of image sharpness and blurring is apparent in the resampled image.Because the baseline R1 maps had less contrast than the post-injection images, it was would be preferable to register, and thus blur, the baseline image to the post-injection image, in order to minimize the impact of the blurring. However, it is also useful to have a consistent target position given by the baseline image when comparing its difference with multiple post-injection images. As well, in some cases the baseline or post-injection image position of the anatomy was more or less appealing, such as being closer to being aligned to the image edge axis, making it preferable to use aparticular image as a target for registration. Due to this mix of factors, which of the baseline or post injection images were registered (and blurred) varied among these images.7.6.2 R1 Difference MapsAfter coregistration of baseline and post-injection R1 maps, and subtracting the baseline from the post injection, R1 difference maps were produced for the rats that received mixed Mn-52 and non-radioactive MnCl2 injections and MR scanning as discussed above. Slices from these maps are shown in figures 7.25 through 7.27.207Figure 7.25: Relaxation rate change map between 4 days post ICV injection and baseline in rat that receivedsuccessful right-later ventricle injection. Scale has units s-1. Largest increases are seen inthe colliculus, and in the right lateral ventricle near the injection site.Figure 7.26: Relaxation rate change map between 8 days post ICV injection and baseline in rat that receivedmisplaced right-later ventricle targeted injection. A distinct pattern with R1 change largenear the injection site but less elsewhere is seen, along with decreases near the lateralventricles. The scale (units of s-1) has negative values in this image, to illustrate decreased R1.Figure 7.27: Relaxation rate change map between 4 days post IP injection and baseline in rat. Scale has units s-1.The olfactory bulbs again appear bright, along with the midbrain and cerebellum.2087.7 DiscussionThe R1 maps shown in figures 7.15 and 7.23 appear to be of qualitatively similar quality to those shown by Chaung (2006). Chaung's images used injections into the olfactory bulb of rats, with volume 50 nl at concentration 100 mM, however, and are although those images presented in transaxial orientation, making direct comparison to coronal images after ICV injections shown here difficult.As noted in the Mixed ICV Injections section above (see page 195), change in R1 was roughly proportional to amount of Mn successfully injected ICV in these experiments, as estimated in different animals in different areas of the brain. More quantitative validation of this proportionality, and the applicability of a calibration factor to convert change in R1 to a change in Mn concentration, are discussed in the Multimodality Comparisons chapter, using autoradiography and PET of the sameanimal as standards for comparison. The images in figures 7.25 through 7.27 provide a useful variety of Mn distributions for these comparisons, including widespread accumulation after systemic and successful ICV injection, and highly localized accumulation after a misplaced ICV injection.As far as is known, maps of change in R1, such as in figures 7.25 through 7.27, have not been previously published prior to this work. Similar relaxivities were reported by Tambalo (2009): near 0.7 baseline, and near 1.0 after IP administration of 0.2 mmol/kg in multiple injections. These resultswere reported only for ROI averages on various brain structures, however. More important for this work, however, is the noted potential for comparison with other modalities to validate the technique.2098 Multimodality ComparisonsSeveral goals for this work involve the use of multiple modalities to image manganese distribution in vivo, and to compare results between the modalities of Mn-52 positron emission tomography (PET) and autoradiography (AR), as well as manganese-enhanced magnetic resonance imaging (MEMRI). As well, the ability of PET to image Mn-52 with and without large doses of Mn, which are required with systemic administration for MEMRI, allows investigating of the effect of such large doses, whichis a potentially important result for planning and interpretation of MEMRI studies. These comparisons are discussed in this chapter.8.1 Multimodality CorrelationAs discussed in the Manganese-Enhanced Magnetic Resonance Imaging Mn Concentration Mapping section (see page 174), estimation of in vivo Mn concentration changes using MRI as done in this work require an assumption that changes in R1 relaxation rate are linearly related to change in Mn concentration in vivo. However, it was expected that this assumption would not hold in all cases, particularly at locally very high Mn concentrations. As discussed in the MR Imaging Mixed ICV injection section (see page 195), changes in MR signal and R1 were observed, after Mn injections, that were locally inconsistent with the assumption, seemingly due to such locally high concentration of Mn, or edema related to its administration. These images alone could not address the question of whether the assumption was applicable throughout the rest of the brain where concentrations wereless extreme and where there was no apparent edema.It is thus a goal of this work to use PET or autoradiography, which are considered to be quantitativelyreliable, to evaluate the degree to which MRI R1 relaxation rate change maps are linearly dependenton the actual concentration of Mn in tissues being imaged. Because MEMRI is used most commonly in the brain of small animal, it was specifically desired to investigate this issue with rats. Phantom experiments would not be suitable to address this issue, as they lack the variety of Mn accumulationpatterns present after accumulation in live brain tissue, and would not feature the same potential complications to the relaxivity linearity assumption (as discussed further in the Manganese-210Enhanced Magnetic Resonance Imaging section, see page 174).To that end, animals were imaged with PET, MRI, and AR after receiving single injections containing both Mn-52 and non-radioactive MnCl2. These injections and the resulting images are discussed in the MRI Imaging, PET Imaging, and Autoradiography sections (see pages 194, 99, and 111). The animals that were used for these comparisons were the two rats that received mixed injections by ICV targeting the right lateral ventricle (of which one was successful and one was misplaced), and the rat that received a mixed IP injection. Comparisons involved coregistration of the separately acquired PET, AR, and MRI images, and then examining scatter plots and correlations between modalities.8.1.1 RegistrationIn order to directly correlate Mn images acquired with PET, MRI, and AR, it is necessary that they have the same number and size of pixels, and the same anatomical structures appearing in the samelocations. Because the images were separately acquired by completely independent means, these conditions were not initially met, and both position and pixel sizes were generally inconsistent. It was thus necessary to coregister images and resample them to be consistent in these regards.The method used for coregistration and resampling was the Fusion tool in the microPET ASIPro software, a screenshot of which is shown in figure 8.1. After initially processing the image data, as discussed in the MRI and AR sections (pages 205 and 114), registration involved loading the separatemodality 3D images into ASIPro, and manually specifying 3D translation and rotations so that matching structures appeared at the same locations in the combined image view.211Figure 8.1: Screenshot captured during coregistration of Mn-52 autoradiograph (top row) to R1 difference image(bottom row) of rat brain using the ASIPro software Fusion tool. This tool allows images to be overlaidat varying opacity (centre row) to visually judge coregistration quality. Rotation and translations in threedimensions may be specified using the Fusion tool transformation interface (not shown).As discussed in the Autoradiography section (see page 125), AR images tend to have spatial distortions and damage, which R1 images do not, which prevented simultaneous coregistration of 212the full volume. Instead, translation and rotation were adjusted until a single slice from each modality appeared registered. Even with this limited form of registration, there was still some deviation between the apparent locations of structures in slice images, but the gross features such as the locations of areas of high accumulation of Mn or R1 difference could be positioned closer than would be possible with full 3D images. A quantitative assessment of registration quality was notattempted, as there is no clear metric for doing so beyond visual inspection, and the relatively small number of images being coregistered was practical to handle manually, without need for an automatic registration method.After coregistration, the AR images were resampled using the translation and rotation, onto a grid that matched the voxels of the MR image.8.1.2 ComparisonTo test the linearity of R1 difference to AR counts in corresponding pixels, scatter plots of the values in the same pixels of a single slice of the coregistered images were generated. AR images were preferred over PET for validation of R1 difference linearity due to their better ability to resolve details of the Mn-52 distribution; as noted previously, the resolutions of Mn-52 PET and AR images are approximately 1.8 mm and 0.47 mm, while the R1 map pixel size, and thus resolution, in this work was 0.3125 mm. After coregistration, the same single slice in both of the 3D volumes for each animal were chosen. The chosen slices had interesting structures and relatively wide ranges of apparent Mn concentration. For example, the peak near the injection site of the animal that received a misplaced injection was included, and all selected slices included part of the colliculus, which appeared frequently in images as a location of Mn accumulation.After selecting slices, thresholding was applied to limit the voxels processed further to those that corresponded to brain tissue, and to exclude surrounding regions. In the MR images, the surroundings were other tissues that are present outside the brain. In the AR images, the background of the plate images surrounded the tissue slices, and appeared as lower-valued pixels. 213Accordingly, no meaningful relationship was expected between those voxels, and comparison of their values was not done.The thresholding was applied based on AR image voxel values, as these provided a clear contrast between tissue and background, whereas the MR images which frequently had comparable voxel values within and surrounding the brain. Accordingly, only voxels that were above a threshold in the AR image, which corresponded to the transition between tissue and background, were included in subsequent scatter plots. Pixels outside the brain tissue have no useful information in the AR data, and are thus not useful to correlate with corresponding MR image pixels. Slice images after applying the thresholds are seen in figure 8.2.Figure 8.2: Slices of R1 difference map (left, in units of s-1) and autoradiography image (right, in digital light units) of rat that receivedmisplaced ICV injection of Mn-52 and MnCl2 after coregistration and applying a threshold filter.Scatter plots of corresponding pixels' Mn-52 AR and MRI R1 difference values were generated, and are shown in the figures 8.3 through 8.5.214Figure 8.3: Scatter plot of change in R1 between baseline and post-injection R1 mapsin rat that received ICV injection of Mn-52 and MnCl2 into the right lateralventricle against the autoradiography counts in corresponding pixels aftercoregistration of single slice brain images of both modalities.215Figure 8.4: Scatter plots of change in R1 between baseline and post-injection R1 maps in rat that receivedmisplaced ICV injection of Mn-52 and MnCl2 near the right lateral ventricle against the autoradiography countsin corresponding pixels after coregistration of single slice brain images of both modalities. These plots show thesame data and fit, but the lower plot has a restricted scale to focus on the region with the majority of data points.216Figure 8.5: Scatter plot of change in R1 between baseline and post-injection R1 mapsin rat that received IP injection of Mn-52 and MnCl2 against the autoradiography counts incorresponding pixels after coregistration of a single slice brain image of both modalities.Two-parameter least squares linear (slope and intercept) functions were fit to these data, as shown by the red lines in the scatter plot figures above. Slope with standard error, intercept with standard error, R2, and parameter t-scores (using null hypotheses of 0 slope and 0 intercept) from these fits are shown in table 8.1. The absolute values of the slope are not important, as they will depend on the timing of the autoradiography acquisition (i.e. the delay between scans and resulting radioactive decay, and the duration of AR acquisition), and are thus not usefully comparable between subjects. The t-scores are all highly significant for > 2000 degrees of freedom, with vanishingly small probability (p << 0.0001) of being observed by random chance if the null hypotheses are true.Injection R2 DataPointsSlope(s-1/(103 counts))Slope TScoreIntercept(s-1)InterceptT ScoreICV successful 0.458 3086 0.0552 +/- 0.0011 51.0 -0.031 +/- 0.004 -7.12ICV misplaced 0.608 2325 0.0487 +/- 0.0008 60.0 0.061 +/- 0.003 19.6IP 0.231 3070 0.2281 +/- 0.0075 30.4 0.023 +/- 0.003 6.95Table 8.1: Linear regression fit statistics for R1 change plotted against autoradiography counts.217For the animal with the misplaced injection, there is a large region of sparsely sampled points above 5000 AR counts with little discernible trend, which likely corresponds to the peak region of the slice. It is unsurprising that correlation in the region is poor, as the concentration of Mn in this case lead tosome artifacts in MR images, and may have reached the threshold where signal loss to R2 relaxation effects prevented adequate quality MR data for R1 measurement from being acquired from those regions to accurately reflect the linearity of concentration that was present at lower concentrations. A second scatter plot is shown for this animal in figure 8.4 (at bottom), with a restricted scale view ofthe same data, below 5000 AR counts, which is similar to the domain of AR seen in the animal with the successful ICV injection. In this plot, a trend similar to that of the ICV injected animal can be seen.For the slice from the IP injected animal the large amount of scatter in the data make meaningful interpretation difficult. This is not surprising, given the relatively small amounts of Mn that were observed to cross the blood-brain barrier in PET images, and the poor statistical quality of the AR image data for this animal.8.2 Multimodality CalibrationAs noted in the previous section, a linear relationship between change in Mn concentration and change in R1 relaxation rate measured with MRI may be assumed, and was indeed observed in someof the tested brain slices by comparison with autoradiographs. Further, as discussed in the Manganese-Enhanced Magnetic Resonance Imaging Relaxivity Calibration section (see page 174), a slope for this relationship may be determined using in vitro measurements. Using this slope, a measured change in R1 may be calibrated to estimate a change in Mn concentration. However, the degree to which such a calibration line, measured in vitro, will accurately reproduce Mn concentrations is unclear.It was desired to check the calibration accuracy by comparison of Mn concentrations measured with MRI and PET. AR was not suitable for this purpose, as in this work AR images are not calibrated to give concentrations of Mn-52 present in the tissues. PET images, however, are calibrated as discussed in the PET Imaging Calibration Phantom section (see page 74), and have voxel values in 218units of radioactivity concentration (e.g. kBq/cc).Calibration factors were determined to convert changes in R1 to Mn concentration, between baseline and post-injection R1 maps generated with MRI, using in vitro non-imaging inversion recovery as discussed in the Manganese-Enhanced Magnetic Resonance Imaging Relaxivity Calibration section (see page 174). The rat in these experiments was maintained at approximately 37.7 C during the R1 mapping scans, as measured with a rectal temperature probe. Accordingly, the calibration slope determined from in vitro experiments for that temperature were used: 4.2 s-1mM-1. Using this calibration factor, the 4 day post-injection R1 difference image for the rat that receive a successful ICV injection of 6.5 mM MnCl2 and Mn-52 was converted to an image of estimated Mn concentration.Using the registration procedure described above, the 5-day post-injection PET image for the same animal was resliced to match the position of the R1 difference image. Single slice images from the PET, MR, and AR data are shown in figure 8.6. The MR concentration image (bottom row) is shown without (right) and after (left) applying a 2 mm FWHM 3D Gaussian kernel by convolution to smooththe resolution to be similar to the PET image, to which it is being compared, and to remove any high spatial frequency noise contributions. These images are also shown after applying the same thresholding procedure described above, based on AR image pixel values. For the MR concentration images and PET image, the pixels included by the threshold were scatter plotted, as shown in figure 8.7.219Figure 8.6: Coregistered single slices from images of a rat that received a right lateral ventricle injection of Mn-52and non-radioactive MnCl2. At top left: PET image. At top right: AR image. Bottom right: Mn concentrationimage derived from MRI R1 relaxation rate change. Bottom left: same as bottom right, but smoothed with a2 mm FWHM kernel to give the MR-derived concentration map the same resolution as the PET image.220Figure 8.7: Scatter plots in estimated concentrations of Mn or Mn-52 from MRI and PETin images of the same rat brain slice after coregistration between modalities.MRI concentration is calculated by applying a calibration factor of 4.2 s-1mM-1 to thechange in R1 relaxation rate between baseline and post-injection images. At top, MR data isplotted unaltered. At bottom, MR data has been smoothed with a 2 mm 3D Gaussiankernel to make its resolution similar to the PET image.221As with the plots of AR against MRI R1 difference, a distinct linear trend can be seen between the MRI Mn concentration estimate and the PET radioactivity concentration. For the unsmoothed MR concentration data, the two-parameter least-squares regression fit line has slope and standard error 0.001457 +/- 0.000028 (mmol/l)/(kBq/cc) and intercept and standard error 0.0060 +/- 0.0008 (mmol/l) with 3084 degrees of freedom, a slope and intercept t-scores of 51.9 and 7.76 with null hypothesis of 0 slope and intercept, and a fit R2 of 0.466. As with the R1 change regression against AR counts discussed above, due to the large number of degrees of freedom, this fit is extremely statistically significant, with a vanishingly small probability (p << 0.0001) of being observed if the nullhypotheses were true. For the smoothed data, the fit line has slope 0.001465 (mmol/l)/(kBq/cc) and intercept 0.0055 (mmol/l), and a fit R2 of 0.912, although correlations between smoothed data points make statistical significance tests difficult.In order to test the accuracy of this slope, an independent estimate of the ratio of Mn concentration to Mn-52 concentration was needed. This was derived from the concentrations of activity and Mn in the ICV-injected solution, accounting for radioactive decay. The initial injection contained approximately 6.5 mM MnCl2 and 9.7 MBq/cc Mn-52 in 30 ?l solution. The nominal 5-day post-injection image was acquired approximately 5.27 days after the injection, which with the Mn-52 half life of 5.6 days gives a decay factor of 0.52, and a decayed Mn-52 activity concentration of 5.04 MBq/cc. The ratio of the Mn molarity to this Mn-52 concentration is 0.00129 (mmol/l)/(kBq/cc). Thisratio is (surprisingly) consistent with the fit slopes near 0.00146 (mmol/l)/(kBq/cc) as reported above; the difference is 13% of the ratio expected from what was injected.Notably, the MRI image was acquired 4 days after injection, while the PET image was acquired 5 daysafter injection. This timing was not ideal for a direct comparison, but was used for practical reasons. The approximate one day difference, during which Mn is likely clearing from the brain (as suggested by MR signal decreases seen after systemic injection (Aoki, 2004) and after ICV injections discussed in the MR Imaging section of this work, see page 196), would lead to a relative increase in the amount of Mn as seen by MR compared with PET, which would lead to a larger slope being estimated.222Additionally, as discussed in the Manganese-Enhanced Magnetic Resonance Other Calibration FactorLimitations section (see page 179), various limitations to the applicability of an in vitro calibration factor as discussed above are expected. Likely the R1 relaxation rate reported in regions of very high Mn uptake are not reliable, and may have some influence on the accuracy of the slope determined from the above scatter plot. Mn concentrations up to approximately 0.1 mM, or related R1 relaxation rate changes of 0.4 s-1 appear to be consistent between modalities, when other factors affecting R1, such as edema, are not present. The number of pixels in the image slices used for this test that exceed these limits is relatively small (i.e. in the 4th ventricle between the colliculi). In the successfully ICV injected animal, Mn was observed to spread throughout the brain, and did not accumulate as dominantly in one location as occurred at the site of the misplaced ICV injection.8.3 Mn Dose EffectsAs discussed in the Manganese-Enhanced Magnetic Resonance Imaging Manganese Toxicity section (see page 173), Mn injections in vivo can have severe toxic effects, or may be well tolerated, depending on the amount given and route of administration. During this work, complications after Mn injections in rats were observed, particularly when giving large doses systemically. There may also be dose-dependent effects of systemic Mn administration that are not readily observed throughacute health effects.Study of such effects is beyond the scope of this work, however the ability to image animals with PET after injections of Mn-52 with or without additional non-radioactive MnCl2 provides an opportunity to study dose-dependent effects. Whether, and what, differences occur in Mn distribution between low and high dose administrations is of particular interest because MEMRI in the brain after systemic administration cannot be done without giving relatively large doses of Mn. Itis useful to know whether these amounts of Mn will substantially influence the pattern of its uptake,as if the low and high dose patterns are different, it may limit the use of MEMRI for some investigations where the patterns differ.It was thus desired to investigate whether the amount of Mn given during a systemic injection wouldmeasurably change the amount or pattern of its uptake as seen with imaging in the brain, and 223elsewhere in the body.Additional motivation for this investigation came about when the initial low-dose systemic Mn-52 PET images were acquired, which showed little to no uptake of Mn-52 in the brain of the rats after IVor IP injections. This was surprising, because T1-weighted MR images of rats after IP or IV injections of MnCl2 do show distinct changes in brain contrast and signal enhancement, which is attributed to that Mn accumulation in the brain.It was hypothesized that this discrepancy might be related to the different doses of Mn given in these experiments. Mn-52 PET was initially conducted with systemic injections containing no added MnCl2, which, as discussed in the Tracer Production Specific Activity section (see page 45), involves injections of less than 1 ?g of Mn. MnCl2 injections for MEMRI involved amounts of 40 mg/kg, or roughly 20 mg per systemic injection. The much larger dose of Mn given for MEMRI could influence the rate of uptake or clearance from the blood, such as if the capacity for the liver to filter Mn from the blood is overwhelmed after a 20 mg dose, which might leave Mn in the blood longer, giving it more time to cross the BBB.Accordingly, a 490 g rat was given a 2.05 ml IP injection containing both 15.1 MBq Mn-52 and 75 mM non-radioactive MnCl2, as discussed in the PET Imaging Animal Mixed Mn-52 and Nonradioactive MnCl2 Imaging section (see page 99). Images were acquired of both the head and abdomen, and will be discussed separately.Images of the heads of two rats that received systemic Mn-52 injections are shown in figures 8.8 and8.9. In these images, regions of interest (ROIs) can be seen. These were drawn to cover the apparent volume of the brain in the PET image, in 24 image planes covering approximately 6300 voxels each. The purpose of these ROIs was to assess the amount of Mn-52 that accumulated within the brain, soas to better quantify any effect from the larger total dose of Mn in the second rat. Also shown in these images are ROIs drawn on the pituitary gland, immediately inferior to the middle of the brain. The pituitary is outside the blood brain barrier, and thus can have a large accumulation of Mn even when little accumulates in the brain itself. The pituitary ROIs covered 3 image planes and approximately 90 voxels, centred on the highest activity planes and points of that structure.224Figure 8.8: Transaxial (left), coronal (centre), and sagittal (right) slices through PET image ofrat head 1 day after IV injection of Mn-52 with high specific activity. Image has been smoothedwith 2 mm FWHM kernel to reduce noise. ROIs covering the pituitary gland (?pit?) and brain(?brain?) are seen as yellow dashed lines in all slice orientations.Figure 8.9: Transaxial (left), coronal (centre), and sagittal (right) slices through PET image ofrat head 1 day after IP injection of Mn-52 and non-radioactive MnCl2. Image has beensmoothed with 2 mm FWHM kernel to reduce noise. ROIs covering the pituitary gland (?pit?)and brain (?brain?) are seen as yellow dashed lines in all slice orientations.Averages of the concentrations within these ROIs were extracted. Given the doses of Mn-52 and body masses of the rats, the standardized uptake values (SUV) for these ROIs were also calculated, 225which are the concentration in the ROI divided by (the radiation dose divided by the body mass). These SUVs are shown in table 8.2.12 MBq Mn-52IV326 g15 MBq Mn-52IP490 g150 ?mol MnCl2Brain ROI mean (kBq/cc) 4.37 2.49SUV ((kBq/cc)/(MBq/g)) 119 81Pituitary ROI mean (kBq/cc) 104.13 29.28SUV ((kBq/cc)/(MBq/g)) 2829 950Table 8.2: Concentration and specific uptake values in PET images in ROIs on brainand pituitary glands of rats after systemic injection of Mn-52 with or withoutadditional MnCl2 mixed into the injection solution.In the head images, the distributions are largely similar. The brain appears as a region of low uptake, likely due to the blood brain barrier, surrounded by a ring of activity that may be the bones of the skull. As corroborated by the concentration and SUV values in brain ROIs, there was relatively little change in the amount of Mn that accumulated in the brain between the high and low Mn dose injections. In the pituitary, conversely, there was a threefold increase in SUV when the Mn dose is low. This suggests that the presence of additional Mn in the blood impairs the relative uptake into the pituitary (although the absolute amount accumulating therein may be higher with the higher dose of Mn).Images of the abdomens of two rats that received systemic Mn-52 injections are shown 8.10 and 8.11.226Figure 8.10: Transaxial (left), coronal (middle) and sagittal (right) slices of PET image of ratabdomen brain 1 day after IV injection of Mn-52. PET emission data (red-white) areoverlaid over attenuation map image (grey) for anatomical context.Figure 8.11: Transaxial (left), coronal (middle) and sagittal (right) slices of PET image of rat abdomen brain1 day after IP injection of Mn-52 and 75 mM MnCl2 in approximately 2 ml volume. PET emission data(red-white) are overlaid over attenuation map image (grey) for some anatomical context.As previously discussed, in the IV injected animal, the pelvic bones and spinal column are clearly visible. In the IP injected animal with additional Mn content, there is little or no discernible structureto the uptake. Blobs of activity are seen throughout the abdomen in both images, although in the IV injected image, these appear to have some structure that may indicate abdominal organs or the digestive tract, while in the IP injected animal, there is less connectivity between the areas of accumulation. The lack of prominent spinal column uptake in images of the IP-injected animal may indicate that the larger dose of Mn given with the mixed injection slows the relative rate of uptake ofMn into bones after systemic administration. The different administration methods, IP vs. IV, may contribute to the different result as well. However, as noted in the PET Imaging Animal Mn-52 Imaging General section (see page 82), Kuo (2005) reported little short-term difference in the accumulation of Mn in the brain after IP and IV administrations, which suggests similar patterns may 227also be expected in the abdomen.8.4 DiscussionThe slope derived from the MRI and PET concentrations plot, and the expected slope given the injected Mn-52 activity and concentration of Mn, agree within 13% of the expected value. This is considered to be excellent agreement, given variety of considerations that may affect it, including delays between acquisitions, apparent breakdown of the relaxivity concentration relationship seen in MR some images, and potential differences in that dependence between in vitro and in vivo environments.Particularly for the successful ICV injected animal, the MRI to AR scatter plot in figure 8.3 similarly shows excellent linear dependence of relaxation rate change with Mn concentration within brain tissue.Increasing the total mass of Mn injected systemically does not appear to increase the fraction of Mn that enters the brain (although the total amount entering the brain may increase). Accordingly, giving large doses of Mn with Mn-52 systemic injections is not an effected method to increase the passage of the Mn-52 through the blood brain barrier.2289 Discussion and ConclusionsThis section describes potential future directions of investigation to extend this work, and presents conclusions.9.1 Future WorkThis work included proof of concept and first results of Mn-52 PET in phantoms and in vivo, as well as mapping of concentration using relaxation rate measured with MRI. There are several future stepsthat could be taken to further develop these techniques.9.1.1 Mn-52 PET With Activation-Induced UptakeThe motivation for developing Mn-52 PET was partly the previously reported studies of activation-induced Mn uptake done with manganese-enhanced MRI, in which localized uptake is observed in the brain due to stimuli such as forepaw stimulation and exposure to sounds in rats or mice. Unlike MRI, in which general brain image contrast enhancement from Mn administration is itself useful, there is little reason to employ Mn PET imaging in the brain without attempting to cause activation-induced uptake, as PET lacks the resolution to see fine details of brain structure in small animals. As such, the potential uses for Mn-52 PET in the brain are dependent on being able to visualize and quantify changes in Mn-52 concentration or uptake due to brain activity or stimuli. Accordingly, a useful next step would be to conduct activation-induced uptake studies in the brain of rats with Mn-52 PET.9.1.2 Blood Brain Barrier DisruptionAs discussed in the Manganese-Enhanced Magnetic Resonance Imaging MEMRI Complications section (see page 172), the blood brain barrier (BBB) inhibits or slows the uptake of Mn into the brain after systemic injection. This is consistent with the lack of uptake seen in the brain after systemic injections of Mn-52 in PET images. However, in order to conduct activation-induced uptake 229experiments without surgical direct brain injections, it would be useful to establish BBB disruption technique with PET imaging of Mn-52. BBB disruption has been demonstrated in published reports, with methods such as mannitol injection or ultrasound, but needs to be integrated into the PET imaging process to be used for Mn-52 PET. BBB disruption would be particularly useful for acute uptake studies, in which short-term stimuli such as forepaw stimulation are used to induce localized Mn accumulation.9.1.3 Bone ImagingAs discussed in the PET Imaging section (see page 88), Mn-52 accumulation was observed in the bones of a rat in PET images acquired after IV injection, and this interpretation was later corroborated by IV injection of F-18 water and PET imaging of the same animal. Fore (1952) suggestsMn accumulation is in cortical bone, but not the marrow. Both F-18 and Tc-99m methylene diphosphonate (MDP; Peller, 1993) are cortical bone tracer used for fracture detection, cancer spread, and bone infections. To asses the utility of Mn-52 for longer-term bone imaging, it would be useful to compare it with Tc-99m MDP, and particularly to examine how distribution changes during fracture healing.9.1.4 Improved Tracer ProductionAs discussed in the Tracer Production section (see page 33) and elsewhere, the methods for production of Mn-52 in Cr foil and for preparation of the Mn-52 into injectable form, while sufficient for this work, could be refined and improved. Any future Mn-52 PET imaging work would benefit from use of such improvements.For Mn-52 production, use of a thinner layer Cr, such as plated Cr on a backing surface that can be placed in the irradiation beam path, would be beneficial. This could reduce the amount of Cr to be separated from produced Mn-52, could reduce the amount of non-radioactive Mn in the tracer solution due to impurities in foil and thus improve specific activity, and could increase the effective energy of the irradiation protons, which would decrease the amount of Mn-54 relative to Mn-52 produced. Similarly, use of higher energy proton irradiation than the 12.5 MeV protons used in this 230work would improve the ratio of Mn-54 to Mn-52, and increase the rate of Mn-52 production. PlatedCr on a suitable backing material might also make handling and preparation of foil prior to irradiation simpler. If plating can be done on site, the cost of materials (such as metal foil) may also be reduced.For radiochemistry, alternative methods should be explored to improve the safety and consistency ofthe tracer solution. As noted in the Tracer Production Mn-52 Radiochemistry section (see page 48), Lahiri (2006) have reported an alternate method for Mn-Cr separation that might be explored. More rigorous quality control and production by more-experienced radiochemists using automated equipment and processes may also be helpful for reducing chances of complications from use of Mn-52 injection.9.1.5 Additional Calibration ComparisonsIn the above text, a comparison of the amounts of Mn-52 and non-radioactive Mn seen by PET and MRI in the same animal imaged one day apart is discussed. This comparison was done in a single animal with one image for each modality. It would be interesting to investigate whether the resultingratio of Mn-52 to Mn between these image is consistent in images of the same animal acquired at different times after injection, and in images of other animals injected with both Mn-52 and non-radioactive Mn by ICV or IP injection.9.1.6 Mn MRI LimitationsAs discussed in the Magnetic Resonance Imaging chapter, a rat that received a misplaced right-lateral ventricle-targeted intra-cerebro-ventricular (ICV) injection of MnCl2 and Mn-52 was observed to have decreased R1 relaxation rate, relative to baseline, in post-injection images, in regions near (but not at) the site of injection. Increased signal intensity in T2-weighted images in the same region were also seen, which was attributed to edema. Conversely, at the site of injection, reduced signal intensity in T2-weighted images was observed, while both bright and dark regions were seen in T1-weighed images. These results suggest that there is a complicated mixture of contributions to signalsat the location of the injection, which are difficult to clearly interpret or isolate.231Most importantly, R1 reduction near the injection site suggests similar contributions may be present at the injection site, making R1 change an unreliable and inaccurate assessment of Mn accumulationin this case. (Similar edema-like R1 reduction were not observed in other animals receiving similar injections). There may also be loss of signal at regions of peak Mn concentration in this animal, due to T2-effects, which limit the reliability of these R1 measurements. Collectively, these results indicatethat there are limits to the applicability of the assumption of linearity between R1 change and change in Mn concentration. However, due to the relatively small volume of these regions in these images, their impact on the correlations shown in the Multimodality Comparisons chapter would be limited. As such, these comparisons are not an effective test of these limitations.In order to plan future Mn imaging experiments, it would be useful to know these limits, however, inorder to be sure than future Mn administrations fall into the regime were R1 relaxation rate change is a reliable measure of Mn accumulation, in order to be able to use MRI-derived Mn accumulation images' full experimental potential. To that end, additional experiments could be conducted to determine the range of Mn concentrations that can be reliably measured by relaxation rate change in rat brain, and to better understand when edema occurs that could bias R1 change-based measurements.9.1.7 Mn MRI Concentration Mapping With Activation-Induced UptakeAs noted in the conclusions and motivation, generating maps of change in relaxation rate or concentration after administration of Mn at a per-voxel level has appealing potential for brain activation studies. Existing methods lack sub-structural detail, and rely on region-of-interest averaged relaxivities, or do not attempt to quantify the amount of Mn accumulation and thus amount of neuronal activation at each voxel. An important future step would thus be to attempt imaging of activation-induced Mn accumulation in the brains of rats, similar to published activation-induced MEMRI studies. Unlike previous work, however, this future work could assess the relative accumulation in different brain voxels, and the relative accumulation in response to different stimuli or different strengths or duration of stimuli.2329.2 ConclusionsThis work features the first Mn-52 positron emission tomography (PET) images in phantoms and in live rats, the first Mn-52 autoradiographs (AR) of rat brain, and magnetic resonance imaging (MRI) derived maps of Mn concentration changes via relaxation rate changes in live rat brain at a per-voxel level. Mn-52 PET tracer was produced by irradiation of Cr foil, and radiochemistry was used to prepare injectable solution of that Mn-52. Comparisons between modalities showed a linear relationship between MRI-derived relaxation rate change or Mn concentration and autoradiographs or PET images of the same animal.Mn-52 PET was demonstrated and its performance characterized in phantoms. Image quality metricsincluding uniform region nose, recovery of activity concentration in small regions, and resolution were similar to the established PET tracer F-18. After applying a cascade background correction algorithm based on rescaling the estimated randoms distribution to match the peripheral sinogram bins of the measured PET data, reconstructed images had background levels under 5%, which was adequately quantitative accuracy for this work. Calibration experiments with two vials of known ratio of concentration in a single PET image accurately reproduced that concentration ratio in reconstructed PET images. Normalization of Mn-52 data worked well with normalization data acquired with F-18 filled phantoms.Mn-52 PET was also demonstrated in vivo in rats for the first time with systemic and direct brain injections. Little Mn-52 was found to accumulate in the brain after systemic injection, even when giving large doses of Mn with PET, despite pronounced contrast enhancement after similar injectionsseen with MRI. This discrepancy is attributed to the efficacy of the blood brain barrier and the background effects limiting sensitivity with PET. Mn-52 was found to accumulate in the bones of rats after systemic injection, and remain visible for weeks after injection, and this pattern was similar to the accumulation seen with F-18 fluoride imaging, which is an established bone imaging tracer. This suggests Mn-52 may be useful as a longer-term bone imaging agent.Mn-52 autoradiography was demonstrated after systemic and direct brain injections of Mn-52 in liverats. The autoradiographs had better ability to resolve small details of the distribution in brain than 233did PET images of the same animals. Due to the lack of calibration data acquired with the images of brain slices, background subtraction was necessary to produce quantitative images. The resulting autoradiographs were useful for correlation with and validation of MRI concentration mapping results.Mn was used as a contrast agent in MRI after systemic and direct brain injections in live rats. Change in longitudinal relaxation rate (R1) was measured by comparing baseline and post-injection R1 maps derived from Look-Locker inversion recovery MRI. Calibration data were acquired using non-imaging inversion recovery, and were used to convert change in relaxation rate to change in Mn concentration. Consistent with previously published results, T1-weighted images in the brain showeddetailed contrast changes, after direct brain injection or systemic injection without blood brain barrier disruption. Corresponding maps of change in MR relaxation rate were produced by coregistering baseline and post-injection relaxation rate maps, and show similarly excellent detail in rat brain.Mn-52 and non-radioactive Mn were administered simultaneously by single IP or ICV injections. PET, AR, and MR images of rats that received these injections were compared. Particularly for a successful right-later-ventricle targeted injection, linear relationships between AR and MRI Mn distributions, and a between PET and MRI Mn distributions were seen. The slope between the PET and MRI-derived Mn concentrations was found to be close to the slope estimated from the known Mn contents of the injection.These results validate the use of MR-derived concentration maps for assessing absolute accumulation and relative distributions in the brain of rats after systemic or direct brain injections. This method has promising potential for increasing the power of Mn MRI uptake studies, which previously relied primarily on determining locations with high likelihood of having Mn uptake, or without assessing amount of uptake a per-voxel level. Mn relaxation rate change and concentration mapping with MRI was found to be potentially powerful tool to improve the experimental power of Mn-uptake imaging to assess neuronal activation in the brain. Mn-52 PET may be less useful for brain imaging due to blood-brain barrier 234uptake and background limitations, but has potential applications outside the brain where MRI is less suitable, including bone imaging.235ReferencesK.C. Allman, "18F-FDG PET and Myocardial Viability Assessment: Trials and Tribulations", The Journal of Nuclear Medicine 51, 505-506 (2010)F. Angenstein, H.G. Niessen, J. Goldschmidt, H. Lison, W.D. Altrock, E.D. Gundelfinger, H. Scheich, "Manganese-Enhanced MRI Reveals Structural and Functional Changes in the Cortex of Bassoon Mutant Mice", Cerebral Cortex 17, 28-36 (2007)I. Aoki, Y. Wu, A.C. Silva, R.M. Lynch, A.P. Koretsky, ?In vivo detection of neuroarchitecture in the rodent brain using manganese-enhanced MRI?, NeuroImage 22, 1046-1059 (2004)M. Aschner, J.L. Aschner, "Manganese Neurotoxicity: Cellular Effects and Blood-Brain Barrier Transport", Neuroscience & Biobehavioral Reviews 15, 333-340 (1991)M. Aschner, M. Gannon, "Manganese (Mn) Transport Across the Rat Blood-Brain Barrier: Saturable and Transferrindependent Transport Mechanisms", Brain Research Bulletin 33, 345-349 (1994)R.D. Badawi, P.K. Marsden, B.F. Cronin, J.L. Sutcliffe, M.N. Maisey, ?Optimizatino of noise-equivalent count rates in 3D PET?, Physics in Medicine and Biology 31, 1755-1776 (1996)B.J. Beattie, R.D. Finn, D.J. Rowland, K.S. Pentlow, "Quantitative imaging of bromine-76 and yttrium-86 with PET: A method for the removal of spurious activity introduced by cascade gamma rays", Medical Physics 30, 2410-2423 (2003)B.A. Berkowitz, R. Roberts, D.J. Goebel, H. Luan, "Noninvasive and Simultaneous Imaging of Layer-Specific Retinal Functional Adaptation by Manganese-Enhanced MRI", Investigative Ophthalmology & Visual Science 47, 2668-2674 (2006)236A. Bertin, J. Steibel, A-I. Michou-Gallani, J-L. Gallani, D. Felder-Flesch, "Development of a Dendritic Manganese-Enhanced Magnetic Resonance Imaging (MEMRI) Contrast Agent: Synthesis, Toxicity (in Vitro) and Relaxivity (in Vitro, in Vivo) Studies", Bioconjugate Chemistry 20, 760?767 (2009)A. Bishop, N. Satyamurthy, G. Bida, G. Hendry, M. Phelps, J.R. Barrio, "Proton Irradiation of [18O]O2: Production of [18F]F2 and [18F]F2 + [18F]OF2", Nuclear Medicine & Biology 23, 189-199 (1996)N.A. Bock, F.F. Paiva, A.C. Silva, "Fractionated manganese-enhanced MRI", NMR in Biomedicine 21, 473-478 (2008)V. Bouilleret, L. Cardamone, C. Liu, et al., ?Confounding Neurodegenerative Effects of Manganese for In Vivo MR Imaging in Rat Models of Brain Insults?, Journal of Magnetic Resonance Imaging 34, 774-784 (2011)M.M. Britton, "Spatial Quantification of Mn2+ and Mn3+ Concentrations in the Mn-Catalyzed 1,4-Cyclohexanedione/Acid/Bromate Reaction Using Magnetic Resonance Imaging", The Journal of Physical Chemistry A 110, 2579-2582 (2006)M.H. Buonocore, "Quantum Mechanical Description of NMR: From Wave Function to Bloch Equation", ISMRM 2013 Presentation (2013)P. Caravan, C.T. Farrar, L. Frullano, R. Uppal, "Influence of molecular parameters and increasing magnetic field strength on relaxivity of gadolinium- and manganese-based T1 contrast agents", Contrast Media and Molecular Imaging 4, 89-100 (2009)D.M. Chauncey Jr., H.R. Schelbert, S.E. Halpern, F. Delano, M.L. McKegney, W.L. Ashburn, and P.L. Hogan, "Tissue Distribution Studies with Radioactive Manganese: A Potential Agent for Myocardial Imaging", Journal of Nuclear Medicine 18, 933-936 (1977)K-H. Chuang, A. Koretsky, "Improved Neuronal Tract Tracing using Manganese Enhanced Magnetic Resonance Imaging with Fast T1 Mapping", Magnetic Resonance in Medicine 55, 604?611 (2006)237L. Davidsson, ?. Cederblad, E. Hageb?, B. L?nnerdal, B. Sandstr?m, ?Intrinsic and Extrinsic Labeling for Studies of Manganese Absorption in Humans?, The Journal of Nutrition 118, 1517-1521 (1988)J.A. Disselhorst, M. Brom, P. Laverman, et al., ?Image-Quality Assessment for Several570 Positron Emitters Using the NEMA NU 4-2008 Standards in the Siemens Inveon Small-Animal PET Scanner?, Journal of Nuclear Medicine 51, 610-617 (2010)P.P. Dmitriev, I.O. Konstantinov, ?Methods for Producing the Mn52 Isotope?, Atomnaya Energiya 26, 467-469 (1968)G. Elizondo, C.J. Fnetz, D.D. Stank, S.M. Rocklage, S.C. Quay, D. Wonah, Y-M. Tsang, M.C-M. Chen, J.T. Ferrucci, "Preclinical Evaluation of MnDPDP: New Paramagnetic Hepatobiliary Contrast Agent for MRImaging", Radiology 178, 73-78 (1991)O. Eschenko, S. Canals, I. Simanova, N.K. Logothetis, "Behavioral, electrophysiological and histopathological consequences of systemic manganese administration in MEMRI", Magnetic Resonance Imaging 28, 1165?1174 (2010)E. Even-Sapir, E. Mishani, G. Flusser, U. Metser, ?18F-Fluoride Positron Emission Tomography and Positron Emission Tomography/Computed Tomography?, Seminars in Nuclear Medicine 37, 462-469 (2007)H. Fore, R.A. Morton, "The Manganese in Bone", The Journal of Biological Chemistry 51,  598-600 (1952)G. Gianutsos, G.R. Morrow, J.B. Morris, "Accumulation of Manganese in Rat Brain Following Intranasal Administration", Fundamental and Applied Toxicology 37, 102-105 (1997)J.C. Gore, "From Bloch Equation to MR Contrasts: Relaxation and Physical Bases of Tissue Contrast",  ISMRM 2013 Presentation (2013)238E. Hess, S. Tak?cs, B. Scholten, F. T?rk?nyi, H.H. Coenen, S.M. Qaim, "Excitation function of the 18O(p,n)18F nuclear reaction from treshold up to 30 MeV", Radiochimica Acta 89, 357-362 (2001)W. Hillen, U. Schiebel, T. Zaengel, ?Imaging performance of a digital storage phosphor system?, Medical Physics 14, 744-751 (1987)G.P. Howles, K.F. Bing, Y. Qi, S.J. Rosenzweig, K.R. Nightingale, G.A. Johnson, "Contrast-Enhanced In Vivo Magnetic Resonance Microscopy of the Mouse Brain Enabled by Noninvasive Opening of the Blood-Brain Barrier With Ultrasound", Magnetic Resonance in Medicine 64, 995?1004 (2010)R.B. Innis, V.J. Cunningham, J. Delforge, M. Fujita, A. Gjedde, R.N. Gunn, J. Holden, S. Houle, S-C. Huang, M. Ichise, H. Iida, H. Ito, Y. Kimura, R.A. Koeppe, G.M. Knudsen, J. Knuuti, A.A. Lammertsma, M. Laruelle, J. Logan, R.P. Maguire, M.A. Mintun, E.D. Morris, R. Parsey, J.C. Price, M. Slifstein, V. Sossi, T. Suhara, J.R. Votaw, D.F. Wong, R.E. Carson, "Consensus nomenclature for in vivo imaging of reversibly binding radioligands", Journal of Cerebral Blood Flow & Metabolism 27, 1533?1539 (2007)J.S. Kim, J.S. Lee, K.C. Im, et al., "Performance Measurement of the microPET Focus 120 Scanner", Journal of Nuclear Medicine 48, 1527-1535 (2007)A.T.J. Klein, F. R?sch, H.H. Coenen, S.M. Qaim, ?Labelling of manganese-based magnetic resonance imaging (MRI) contrast agents with the positron emitter 51Mn, as exemplified by manganese-tetraphenyl-porphin-sulfonate (MnTPPS4)?, Applied Radiation and Isotopes 62, 711?720 (2005)J.S. Kim, J.S. Lee, K.C. Im, et al., ?Performance Measurement of the microPET Focus 120 Scanner?, Journal of Nuclear Medicine 48, 1527-1535 (2007)M.Z. K?yl?, S. Asubay, A. Yilmaz, "Determination of Proton Relaxivities of Mn(II), Cu(II) andCr(III) added to Solutions of Serum Proteins", Molecules 14, 1537-1545 (2009)239Y-T. Kuo, A.H. Herlihy, P-W. So, K.K. Bhakoo, J.D. Bell, "In Vivo Measurements of T1 Relaxation Times in Mouse Brain Associated with Different Modes of Systemic Administration of Manganese Chloride", Journal of Magnetic Resonance Imaging 21, 334?339 (2005)R. Laforest, X. Liu, "Cascade removal and microPET imaging with 76Br", Physics in Medicine and Biology 54, 1503-1531 (2009)S. Lahiri, D. Nayak, G. Korschinek, ?Separation of No-Carrier-Added 52Mn from Bulk Chromium: A Simulation Study for Accelerator Mass Spectrometry Measurement of 53Mn?, Analytical Chemistry 78, 7517-7521 (2006)L E. Larsen, D. Grant, "General Toxicology of MnDPDP", Acta Radiologica 38, 776-779 (1997)C. Le Loirec, C. Champion, ?Track structure simulation for positron emitters of physical interest. Part I: The case of the allowed decay isotopes? Nucl. Instrum. Methods Phys. Res. A 582, 644?653 (2007).C. Le Loirec, C. Champion, ?Track structure simulation for positron emitters of physical interest. Part II: The case of the radiometals? Nucl. Instrum. Methods Phys. Res. A 582, 654?664 (2007).J.L. Lear, ?Principles of Single and Multiple Radionuclide Autoradiography? chapter in book ?Positron Emission Tomography and Autoradiography Principles and Applications for the Brain and Heart?, Raven Press (1986)Y. Lin, A.P. Koretsky, ?Manganese Ion Enhances TI-Weighted MRI during Brain Activation: An Approach to Direct Imaging of Brain Function?, Magnetic Resonance in Medicine 38, 378-388 (1997)H. Lu, Z-X. Xi, L. Gitajn, W. Rea, Y. Yang, E.A. Stein, "Cocaine-induced brain activation detected by dynamic manganese-enhanced magnetic resonance imaging (MEMRI)", Proceedings of the National Academy of Sciences of the United States of America 104, 2489?2494 (2007)S. Matej, J.S. Karp, R.M. Lewitt, A.J. Becher, "Performance of the Fourier rebinning algorithm for PET with large acceptance angles", Physics in Medicine and Biology 43, 787?795 (1998)240C.L. Melcher, "Scintillation Crystals for PET", The Journal of Nuclear Medicine 41, 1051-1055 (2000)A. Merbach, L. Helm, ?. T?th (Eds.), "The Chemistry of Contrast Agents inMedical Magnetic Resonance Imaging Second Edition", Book, John Wiley & Sons, Ltd. (2013)P. Mi?ville, P. Ahuja, R. Sarkar, S. Jannin, P.R. Vasos, S. Gerber-Lemaire, M. Mishkovsky, A. Comment, R. Gruetter, O. Ouari, P. Tordo, G. Bodenhaus, "Scavenging Free Radicals To Preserve Enhancement and Extend Relaxation Times in NMR using Dynamic Nuclear Polarization", Angewandte Chemie International Edition 49, 6182-6185 (2010)Y. Murayamaa, B. Webera, K.S. Saleemb, M. Augatha, N.K. Logothetis, ?Tracing neural circuits in vivo with Mn-enhanced MRI?, Magnetic Resonance Imaging 24, 349?358 (2006) NEMA - National Electrical Manufacturers Association, "NEMA Standards Publication NU 4 ? 2008 Performance Measurements of Small Animal Positron Emission Tomographs", (2008)NIST compos -  National Institute of Standards and Technology Materials Composition database:http://physics.nist.gov/cgi-bin/Star/compos.plAccessed 12 September, 2013NIST PSTAR - National Institute of Standards and Technology PSTAR database: http://physics.nist.gov/PhysRefData/Star/Text/PSTAR.htmlAccessed 12 September, 2013NIST XCOM - National Institute of Standards and Technology Photon Cross Sections Database:http://www.nist.gov/pml/data/xcom/index.cfmAccessed 12 September, 2013NNDC EXFOR - National Nuclear Data Center Experimental Nuclear Reaction Database:http://www.nndc.bnl.gov/exfor/exfor.htmAccessed 4 February, 2014241NNDC MIRD - National Nuclear Data Center Medical Internal Radiation Dose database:http://www.nndc.bnl.gov/mird/Accessed 12 September, 2013J.T. Nofiele, H-L.M. Cheng, "Ultrashort Echo Time for Improved Positive-Contrast Manganese-Enhanced MRI of Cancer", PLOS ONE 8, e58617 (2013)P.D. Olcott, J.A. Talcott, C.S. Levin, F. Habte, A.M.K. Foudray, "Compact Readout Electronics for Position Sensitive Photomultiplier Tubes", IEEE Transactions on Nuclear Science 52, 21-27 (2005)R. Pautler, A. Silva, A. Koretsky, ?In vivo neuronal tract tracing using manganese enhanced magnetic resonance imaging?, Magnetic Resonance in Medicine 40, 740-748 (1998)G. Pelled, H. Bergman, T. Ben-Hur, G. Goelman, "Manganese-Enhanced MRI in a Rat Model of Parkinson?s Disease", Journal of Magnetic Resonance Imaging 26, 863?870 (2007)P.J. Peller, V.B. Ho, M.J. Kransdorf, "Extraosseous Tc-99m MDP Uptake: A Patho-physiologic Approach", RadioGraphics 13, 715-734 (1993)I. Pesaresi, M. Cosottini; E. Neri, M. Cosottini, D. Caramella (Eds.) "MR Angiography Contrast Agents" in "MR Angiography of the Body Technique and Clinical Applications", Springer (20010)D.J. Porter, L.W. Raymond, G.D. Anastasio, "Chromium: Friend or Foe?", Archives of Family Medicine 8, 386-390 (1999)G. Pratx, S. Surti, C. Levin, ?Fast List-Mode Reconstruction for Time-of-Flight PET Using Graphics Hardware?, IEEE Transactions on Nuclear Science 58, 105-109 (2011)G.D. Rabinovici, A.J. Furst, J.P. O'Neil, C.A. Racine, E.C. Mormino, S.L. Baker, S. Chetty, P. Patel, T.A. Pagliaro, W.E. Klunk, C.A. Mathis, H.J. Rosen, B.L. Miller, W.J. Jagust, "11C-PIB PET imaging in Alzheimer disease and frontotemporal lobar degeneration", Neurology 68, 1205-1212 (2007)242D.W. Rickey, R. Gordon, W. Huda, "On Lifting the Inherent Limitations of Positron Emission Tomography by Using Magnetic fields (MagPET)", Automedica 14, 355-369 (1992)B.D. Rooney, J.D. Valentine, ?Scintillator Light Yield Nonproportionality: Calculating Photon ResponseUsing Measuring Electron Response?, Transactions on Nuclear Science 44, 509-516 (1997)K. Shibuya, E. Yoshida, F. Nishikido, T. Suzuki, T. Tsuda, N. Inadama, T. Yamaya and H. Murayama, "Limit of Spatial Resolution in FDG-PET due to Annihilation Photon Non-Collinearity", International Federation for Medical and Biological Engineering Proceedings 14, 1667-1671 (2007)A.C. Silva, J.H. Lee, I. Aoki, A.P. Koretsky, "Manganese-enhanced magnetic resonance imaging (MEMRI): methodological and practical considerations", NMR in Biomedicine 17, 532?543 (2004)N. Soppera, E. Dupont, M. bossant, "JANIS Book of proton-induced cross-sections", OECD Nuclear Energy Agency (2012)G. Soria, D. Wiedermann, C. Justicia, P. Ramos-Cabrer, M. Hoehna, "Reproducible imaging of rat corticothalamic pathway by longitudinal manganese-enhanced MRI (L-MEMRI)", NeuroImage 41, 668?674 (2008)V. Sossi, J.E. Holden, G.J. Topping, M-L. Camborde, R.A. Kornelsen, S.E. McCormick, J. Greene, A.R. Studenov, T.J. Ruth, D.J. Doudet, "In vivo measurement of density and affinity of the monoamine vesicular transporter in a unilateral 6-hydroxydopamine rat model of PD", Journal of Cerebral Blood Flow and Metabolism 27, 1407-1415 (2007)J.O. Stoner JO, ?Stretched chromium foils?, Nuclear Instruments and Methods in Physics Research A 480, 44-49 (2002)E.M. Strome, S. Jivan, D.J. Doudet, ?Quantitative in vitro phosphor imaging using [3H] and [18F] radioligands: the effects of chronic desipramine treatment on serotonin 5-HT2 receptors?, Journal of Neuroscience Methods 141, 143-154 (2005)243S.C. Strother, M.E. Casey, E.J. Hoffman, ?Measuring PET Scanner Sensitivity: Relating Countrates to Signal-to-Noise Ratios using Noise Equivalent Counts?, IEEE Transactions on Nuclear Science 37, 783-788 (1990)A. Takeda, "Manganese action in brain function", Brain Research Reviews 41, 79?87 (2003)S. Tambalo, A. Daducci, S. Fiorini, F. Boschi, M. Mariani, M. Marinone, A. Sbarbati, P. Marzola, ?Experimental Protocol for Activation-Induced Manganese-Enhanced MRI (AIM-MRI) Based on Quantitative Determination of Mn Content in Rat Brain by Fast T1 Mapping?, Magnetic Resonance in Medicine 62, 1080?1084 (2009)S. T?trault, O. Chever, A. Sik, F. Amzica, "Opening of the blood-brain barrier during isoflurane anaesthesia", European Journal of Neuroscience 28, 1330-1341 (2008)P.S. Tofts, A. Porchia, R. Roberts, B.A. Berkowitz, "Toward clinical application of manganese-enhanced MRI of retinal function", Brain Research Bulletin 81, 333-338 (2010)J. Tropp, "The Reciprocity Principle in NMR Reception", ISMRM 2013 Presentation (2013)S. Vallabhajosula ?18F-Labeled Positron Emission Tomographic Radiopharmaceuticals in Oncology: An Overview of Radiochemistry and Mechanisms of Tumor Localization?, Seminars in Nuclear Medicine 37, 400-419 (2007)C. Watson, C. Hayden, M. Casey, J. Hamill, B. Bendriem, ?Prompt gamma correction for improved quantification in 82Rb PET?, Journal of Nuclear Medicine 49 (Supplement 1) (2008)C.C. Watson ?New, Faster, Image-Based Scatter Correction for 3D PET?, IEEE Transactions on Nuclear Science 47, 1587-1894 (2000)H.I. West Jr., R.G. Lanier, M.G. Mustafa, ?52Cr(p,n)52Mn g,m and 52Cr(d,2n)52Mn g,m excitation functions?, Physical Review C, 2067-2076 (1987)244F. Yang, S. Horng, Y. Lin, Y. Kao, ?Association Between Contrast-Enhanced MR Images and Blood-BrainBarrier Disruption Following Transcranial Focused Ultrasound?, Journal of Magnetic Resonance Imaging 32, 593-599 (2010)A. Yilmaz, M. Yurdako?, B. I?ik, "Influence of Transition Metal Ions on NMR Proton T1 Relaxation Times of Serum, Blood, and Red Cells", Biological Trace Element Research 67, 187-193 (1999)X. Yu, Y.Z. Wadghiri, D.H. Sanes, D.H. Turnbull, ?In vivo auditory brain mapping in mice with Mn-enhanced MRI?, Nature Neuroscience, 961-968 (2005)X. Yu, J. Zou, J.S. Babb, G. Johnson, D.H. Sanes, D.H. Turnbull, ?Statistical mapping of sound-evoked activity in the mouse auditory midbrain using Mn-enhanced MRI?, Neuroimage 39, 223-230 (2008)S. Zhang, Z. Zhou, J. Fu, "Effect of manganese chloride exposure on liver and brainmitochondria function in rats", Environmental Research 93, 149?157 (2003)245Appendix A - Imaging SubjectsThis appendix lists the Sprague-Dawley rats used for imaging experiments in this work, and the scansacquired of those animals. Rats (12 total) were given injections of MnCl2  (4), Mn-52 (5), or both (3) to act as contrast for magnetic resonance imaging (MRI), or as a source of signal for positron emission tomography (PET) or autoradiography (AR). Injections were given intraperitoneal (IP), intravenous (IV), or intra-cerebro-ventricular (ICV). Images and details of procedures are discussed inthe Positron Emission Tomography Imaging, Autoradiography, and Magnetic Resonance Imaging chapters.1) First IP Mn-52 injection rat: This rat had body mass 359 g, and received an IP injection of 600 kBq of Mn-52 while awake. 1 and 2 days after the injection, PET scans were acquired of this rat's abdomen.2) First IV Mn-52 injection rat: This rat had body mass 432 g, and received an IV injection of 10 MBq Mn-52 in 2 ml saline, given over approximately 5 seconds while anesthetized. The rat died after approximately 5 minutes, after which 4 PET scans were acquired to produce a composite full-body image.3) Second IV Mn-52 injection and F-18 IV injection rat: This rat received two sets of injections and scans. First, whe