UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Quantitative NMR imaging and its applications in vivo Stewart, Wendy Anne 1985-06-30

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata

Download

Media
831-UBC_1986_A6_7 S77.pdf [ 5.14MB ]
Metadata
JSON: 831-1.0059464.json
JSON-LD: 831-1.0059464-ld.json
RDF/XML (Pretty): 831-1.0059464-rdf.xml
RDF/JSON: 831-1.0059464-rdf.json
Turtle: 831-1.0059464-turtle.txt
N-Triples: 831-1.0059464-rdf-ntriples.txt
Original Record: 831-1.0059464-source.json
Full Text
831-1.0059464-fulltext.txt
Citation
831-1.0059464.ris

Full Text

QUANTITATIVE NMR IMAGING AND ITS APPLICATIONS J_N VI t>y WENDY ANNE STEWART B.Sc, University of Dundee, 1982 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE i n THE FACULTY OF GRADUATE STUDIES (Department of Chemistry) We accept this thesis as conforming to the required standard. THE UNIVERSITY OF BRITISH NOVEMBER, 1985 (c) Wendy Anne Stewart, COLUMBIA 1985 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of CIM 4£j\\ The University of British Columbia 2075 Wesbrook Place Vancouver, Canada V6T 1W5 Date \IA ALU 1Q\ ABSTRACT The problem of quantifying NMR parameters, measured at a field strength of 0.15 Tesla using a whole body Imaging Instrument, and their potential use j_n vi vo, have been i nvest i gated. The spin-lattice relaxation times (Tj) of water doped with various concentrations of paramagnetic species were determined using the inversion-recovery method. Intensity was measured directly from images as a function of tau and Tj obtained from a three parameter exponential fit of the data. The effects of varying imaging conditions on the values of Tj obtained, were also examined. The results were compared with Tj values obtained using a two-point computational method which is available on many commercial imaging instruments. This involves taking the ratio of an inversion-recovery image and a spin-echo image, which eliminates the dependence of the images on the equilibrium magnetization and the repeat time. The spin-spin relaxation times (T£) were determined using the spin-echo method, of water doped with the same concentrations of paramagnetic species used to study T^. Intensities were again obtained directly from images as a function of 2 tau and obtained from a two parameter exponential fit of the data. The effects of diffusion on the values of 1^ obtained were also examined. The values were compared with those obtained from i i a two-point computational method, which takes the ratio of two spin-echo images with different tau-times. The Tj and T2 data were also compared with literature values obtained under conventional spectroscopic conditions, with no magnetic field gradients present. The results of these studies, which compare favourably with those in the literature, have shown that it is possible to obtain reproducible values of T in the range 100-600 ms, with acceptable errors (±127.) under variable imaging conditions. Reproducible values of can be obtained in the range 40-200 ms, which have errors of ±15% or less. Above this range the effects of diffusion become important. Experimental allergic encephalomyelitis (EAE), an animal model for multiple sclerosis, was induced in a Macaca  fasc i cu1ar i s monkey, and the development of the disease was followed using quantitative NMR imaging. This technique has been shown to be a powerful tool in the study of EAE in primates, since the progression of the disease is accompanied by changes in Tj and T2• The indications are that these changes will allow discrimination between areas of inflammation and others which contain demye1ination. i i i CONTENTS Page Abstract i i List of Tables vi List of Figures x List of Abbreviations xv Glossary of Terms xvii Acknow 1 edgements x Introduction 1 Chapter i 24 Evaluation of the Instrument 2a. Magnetic Field Inhomogeneity 24 b. Spatial Resolution 28 c. Attenuation 30 d. Phasing and Reconstruction 32 Chapter 2 45 Quantitation of NMR Parameters 4a. Spin-lattice Relaxation ...45 ( i ) Definition 4(ii) Measurement of Tj 48 (iii) Computed T 52 (iv) Effects of a Spin-echo Readout 54 (v) Effects of Proton Density 55 (vi) Multiexponential Relaxation Behaviour...57 i v (vii) Discussion 60 b. Spin-spin Relaxation 2 ( i ) Definition 6(ii) Measurement of 1^ 63 (iii) Computed T"2 67 (iv) Effects of Diffusion 69 (vi ) Discussion 70 Chapter 3 72 Applications of Quantitative NMR Imaging 72 ( i ) Background 7(i i) Induct i on of EAE and NMR Imaging Protocol 74 (i i i) Development of EAE using NMR Imaging 76 (iv) Quantitation of NMR Parameters 79 (v) Discussion 81 Conclusions 84 Future Work 6 References 8 Appendix I 93 v LIST OF TABLES Page Table 1: Intensities from two spin-echo 32 images comparing manual and automatic setting of the attenuation. Intensities are given as the mean values from 2.6 cm2 plus or minus standard deviation, at the centre of the vials. Two values are given for each solution corresponding to two different positions in the receiver coi1. Table II: Tj values for various 51 concentrations of CuS04 and MnCl2 solutions obtained at 20 (± i)°C using the inversion-recovery method. Two values are given for each concentration corresponding to two different positions In the rece1ver co i1. Table III: Computed Tj values for various 53 concentrations of CuSO^ and MnCl£ vi solutions obtained at 20 (± 1)°C. Two values are given for each concentration, corresponding to two different positions in the recei ver coi1. Table IV: Tj values for various 55 concentrations of MnC^ solution, obtained at 20 (± i)°C using the inversion-recovery method, and a spin-echo readout with T = 20 ms. Computed Tj values with the same readout are given for comparison. Two values are given for each concentration, corresponding to two different positions in the' recei ver coi1. Table Va: Tj values for various 56 concentrations of MnCl^ solution, obtained at 20 (± i)°C using the inversion-recovery method, and a slice thickness of 20 mm. Computed Tj values are given for comparison. vi 1 Table Vb: T values for various 57 concentrations of MnCl2 solution, obtained at 20 (± 1)°C using the inversion-recovery method, and a slice thickness of 5 mm. Computed Tj values are given for compar i son. Table VI: T^ values for various 66 concentrations of CuS04 and MnCl2 solution, obtained at 20 (± i)°C using the spin-echo method. Two values are given for each concentration corresponding to two different positions in the rece i ver coi1. Table VII: Computed T2 values for various 68 concentrations of CuSO^ and MnCI2 solutions, obtained at 20 (± 1)°C. Two values are given for each concentration corresponding to two different positions in the recei ver coi1. vi 1 i Table VIII: Computed T2 values for various 69 concentrations of MnCl2 solutions, obtained at 20 (± i)°C using various combinations of -[-times. Two values are given for each concentration corresponding to two different positions in the recei ver coiI. Table IX: Tj values as a function of time 79 after induction of EAE, for white matter (WM) and grey matter (GM) which appear normal, and the 1es ion. Table X: T2 values as a function of time 80 after induction of EAE, for white matter (WM) and grey matter (GM) which appear normal and the 1es ion. Table XI: Tj vaiues from five different 81 slice images immediately prior to death for white matter (WM) and grey matter (GM) which appear normal, and the lesion. ix LIST OF FIGURES Page Figure 1: Single projections for back projection 4 i mag i ng. a. Application of gradient G^ b. Application of gradient Gy c. Application of G plus G Figure 2: Two dimensional Fourier transform pulse 5 and gradient sequence. Figure 3: Slice selection in imaging. 7 Figure 4: Selective radiofrequency pulse in the 8 time domain and its Fourier transform. Figure 5: Vector diagrams demonstrating 11 the effects of an inversion-recovery pulse sequence on a spin-system. Figure 6: Plot of intensity versus tau, the 12 inter-pulse delay, for an inversion-recovery pulse sequence. Figure 7: Spin-echo pulse sequence. Schematic 13 diagrams showing formation of the echo. Figure 8: Plot of intensity versus 2 tau for a 15 spin-echo pulse sequence. Figure 9: Sagittal image of a human head obtained 17 using a spin-echo pulse sequence. x Figure 10: Fourier transform of a single projection 20 in back projection imaging. Figure 11: Fourier transform of a single projection 21 in back projection imaging showing phase errors. Figure 12: Diagram of parallel tube phantom for 25 studying magnetic field inhomogeneity. Figure 13: Free Induction Decay (FID) pulse and 26 gradient sequence. Figure 14: FID image of parallel tube phantom 26 in the xz plane. Figure 15: Plot of intensity as a function 27 of distance along the x-direction. Figure 16: Inversion-recovery pulse and gradient 28 sequence. Figure 17: Inversion-recovery image of capillary 29 tube phantom. Figure 18: Diagram of resolution phantom. 29 Figure 19: Spin-echo image of resolution phantom. 30 Figure 20: Phantom used for studying attenuation 31 and in Tj and T2 measurements. Figure 21: a. Plot of intensity versus tau for 34 • 4.96 mM and O 0.99 mM CuSC>4. b. Plot of intensity versus tau for 34 • 4.96 mM and O 0.99 mM CuS04? with magnitude reconstruction. x i Figure 22: Phantom used for studying intensity 35 ambiguities A = 4.96 mM CuS04 B = 0.99 mM CuSC-4 Figure 23: Inversion-recovery images demonstrating 37 intensity ambiguities. a. X - 50 ms b. X = 100 ms c. X = 150 ms d. X = 250 ms e. X = 300 ms f. X = 400 ms g. X = 500 ms h. X = 600 ms i . X 1000 ms Figure 24: Plot of intensity versus tau showing 38 fluctuations in signal sign. • 4.96 mM CuS04 O 0.99 mM CuS04 Figure 25: a. Inversion-recovery image of phantom 42 used to demonstrate intensity ambiguities at x = 200 ms. b. The same image in a. after 42 magnitude reconstruction. Figure 26: Plot of intensity versus tau showing 59 mu1tiexponentia1 relaxation behaviour. O 2.54 mM CuS04 & 0.99 mM CuS04 X 2.54 mM CuSCK + 0.99 CuSO. 4 4 Figure 27: a. Plot of spin-lattice relaxation rate 60 versus concentration for Cu504 so 1ut i ons. xi i O IR data A Computed data b. Plot of spin-lattice relaxation rate 60 versus concentration for MnC^ solutions. O IR data A Computed data Figure 28: a. Plot of spin-spin relaxation rate 71 versus concentration for CuSO, 4 so 1ut i ons. O SE data A Computed data b. Plot of spin-spin relaxation rate 71 versus concentration for MnCl2 solutions O SE data A Computed data Figure 29: Diagram of nerve cell. 72 Figure 30: Diagram showing the position of the 76 monkey in the instrument and the slices being obtained. Figure 31: Transverse SE image of the monkey's 77 brain, showing the abnormal area in the left hemisphere. (Right side of image). xi i i Figure 32: Series of three spin-echo images 78 obtained from the monkey's brain. a. 16.25 days after inoculation b. 16.75 days after inoculation c. 18.42 days after inoculation The light areas are abnormal. x i v LIST OF ABBREVIATIONS adenosine triphosphate static magnetic field radiofrequency field Bohr magneton cent imetre computed tomography copper(II) sulphate decibels, measure of attenuation degrees celcius static magnetic field inhomogeneity radiofrequency field inhomogeneity grad i ent non-1i near i ty diffusion coefficient experimental allergic encephalomyelitis free induction decay magnetic field gradient along the x-direction magnetic field gradient along the y-directi on magnetic field gradient along the z-directi on grey matter of the brain gamma, gyromagnet i c rat i o hertz nuclear spin intensity in an inversion-recovery image intensity in a spin-echo image xv IR inversion-recovery mM millimo1ar mm millimeter MnCl2 manganese(I I) chloride MS multiple sclerosis MQ equ i1i br i um magnet i zat i on My, magnetization in the xy plane M^ magnetization along the z-directi on v nu, Larmor precession frequency in hertz NMR nuclear magnetic resonance u omega, Larmor precession frequency in radians per sec 3 1 P phosphorous-31 % percent PET positron emission tomography RF radiofrequency p rho, spin density S electron spin T tau, the interpulse delay T. correlation time c Tj spin-lattice relaxation time T^ spin-spin relaxation time TR repeat time, the time between successive reapplication of a pulse sequence 2DFT two dimensional Fourier transformation WM white matter of the brain xvi GLOSSARY OF TERMS Central nervous system: This consists of the brain and spinal cord. Cerebral hemisphere: Either of the pair of structures constituting the main portion of the brain, occupying the upper part of the cranial cavity. Demye1ination: Destruction or removal of the myelin sheath of a nerve or nerves. Encephalomyelitis: Inflammation involving both the brain and spinal cord. „ Gross pathology: Diseased tissues visible to the naked eye. Haemorrhagic necrosis: Death of tissue, usually as individual cells, groups of cells or in small localized areas, due to rupturing of blood vessels. Histology: That department of anatomy, which deals with the minute structure, composition and function of the tissues. Histopathology: the histology of diseased tissue. Inflammation: A localized protective response elicited by injury or destruction of tissue, which serves to destroy, dilute, or wall off both the injurious agent and the injured tissue. Histologically, it involves a complex series of events which include dilation of arterioles, capillaries and venules, with increased permeability and blood flow. Intradermal: Within the dermis, which is the outer layer of skin deep to the epidermis, consisting of a dense bed of vascular connective tissue. Ketamine hydrochloride: Chemical name (±)-2-o-chlorophenyl-2-methy1 aminocyc1ohexanone hydrochloride. A non-barbiturate, rapid acting general anaesthetic which can be administered intravenously or intramuscularly. Mycobacter i um tubercu1osi s: Species of micro-organism which causes tuberculosis. The disease is characterized by the formation of tubercles and caseous necrosis in the t i ssues. Myelin: The lipid substance forming a sheath around certain nerve fibers. Neuron: Any of the conducting cells of the central nervous system. A typical neuron consists of a cell body, containing the nucleus and the surrounding cytoplasm; several short radiating processes known as dendrites, and one long process (axon) which terminates in twig like branches, and may have branches along its course. Rompun: Chemical name, 2-(2,6-dimethylphenylamino)-4H-5,6-dihydro-1,3-thiazine. Potent sedative, hypnotic. Can be administered intravenously or intramuscularly. xv i i i Synapse: The anatomical relation of one nerve cell to another; the region of junction between processes of two adjacent neurons, forming the place where a nervous impulse is transmitted from one neuron to another. xix ACKNOWLEDGEMENTS I would like to thank Professor L.D. Hall for his support and encouragement throughout this work. I would also like to express my appreciation to Dr. D.W. Paty, for his support and enthusiasm throughout our collaborative research. Thanks also go to Mr. D. Aikins, whose engineering expertise has been invaluable, and to the Research and Development group at Picker International, Cleveland, Ohio, for their helpful discussions. I give special thanks to Anneke Rees for typing this thesis, and for her patience and encouragement throughout its preparation. xx Ded i cat i on To Professor Roy Foster XXI INTRODUCTION The phenomenon of nuclear magnetic resonance (NMR) was 1 2 first discovered by Puree 11 and Bioch ' in 1946, and is now an important analytical tool for both scientists and phys i c i ans. NMR spectroscopy has been used by scientists in the elucidation of complex chemical structures, such as steroids and proteins, in reaction product characterization, and in the study of electron-donor-acceptor complexes, to name a 3 few examples from organic chemistry. With progressive technical improvements in NMR instrumentation, many biological studies have also become possible. In 1971, 4 Damadian carried out jjn vitro proton NMR studies on rat tissue, and demonstrated differences between the relaxation behaviour of the protons in normal and cancerous tissue. This provided the first evidence that NMR signals could be used to discriminate between normal and diseased states. In 1973, the application of NMR to medicine was extended when 5 31 Moon and Richards detected the intraerythrocytic P resonances of 2,3-diphosphog1ycerate and inorganic phosphate from blood. They also demonstrated that the intraerythro cyt ic pH could be deduced from a detailed study of the inorganic phosphate chemical shift. In 1974, the work of Moon and Richards was extended by 6 31 Henderson, Costello and Omachi , who also detected the P resonances from adenosine triphosphate (ATP) in human 1 erythrocytes. In the same year, Hoult e_£ aj_. showed that the 31P NMR signals from ATP, phosphocreatine and inorganic phosphate could be obtained from intact muscle samples. ATP is the key metabolite in virtually every energy transfer process in the body. Its regeneration involves both inorganic phosphate and phosphocreatine; therefore metabolism could be followed for the first time j_n v i vo using 31P NMR spectroscopy. In addition, since the technique is non-invasive and does not require ionizing radiation, metabolic disorders could be followed over time 8 9 and the effects of therapy monitored . Over the past decade a large number of J_n v i vo spectroscopic studies have been carried out^ and many more are in progress.'4 Imaging Methods 1. Back Projection In most conventional NMR experiments the sample under study is placed in a uniform static magnetic field, BQ, and the equilibrium magnetization in the z-directi on is perturbed by application of a radiofrequency (RF) pulse at 15 the resonance frequency of the appropriate nucleus . This work is concerned with proton NMR. The resonance frequency is given by equation 1, »o - yBo (1) where y = gyromagnetic ratio of the proton. 2 The ability to obtain spatially encoded information using the NMR signals from a sample was first demonstrated by Lauterbur in 197316. In the imaging experiment, magnetic field gradients are applied, resulting in frequency labelling of the sample with respect to distance along the direction of the applied gradient. If the applied gradient (G for example) is X linear, then equation 1 becomes ca - vB + Y G (2) x ' o x x where is the resonance frequency of nuclei at position x during application of gradient G . X Cons i der a samp1e compr i s i ng two tubes conta i n i ng the same solution (for example, water) placed in a static magnetic field, as shown in Figure 1. If a linear Gx gradient is applied during data acquisition after a 90° RF pulse, then the Fourier transform17 of the time domain signal has the form of the projection shown in Figure la; the water in the two tubes has different resonance frequencies corresponding to the positions of the tubes along the x-axis. Similarly, application of a gradient along the y-axis during data acquisition and subsequent Fourier transformation provides the projection shown in Figure lb. Now the water in both tubes has the same resonance frequency, and a single profile is obtained with twice the intensity from a single tube. If linear 3 combinations of the x- and y-gradients are used, a simple effective linear gradient is obtained, and hence projections may be obtained at any angle with respect to the y-axis (for example, 45°, as shown in Figure lc). Usually, 180 projections are obtained, with the composite gradient %rotated' at 1° intervals. The Fourier transforms of each l ft projection are then "filtered" and "back projection" carried out to provide the final image. y a /N /is Bo * ~) o o Figure 1: Single projections for back projection i mag i ng. a. Application of gradient b. Application of gradient G c. Application of G plus G x y 2. Two-Dimensional Fourier Transformation (2DFT) The idea of using Fourier methods in imaging was introduced by Kumar, Welti and Ernst in 197519. 2DFT imaging is a specific example of a broader class of NMR 20 techniques known as 2DFT spectroscopy . An example of a pulse and gradient sequence used for 2DFT imaging is shown in Figure 2. If the imaging method of plane selection is used, the experiment begins with the application of a 180* RF Slice Select Gradient Read Gradient Phase Encoding. Gradient Data Collect W I «5*» -Jl" 1 —~~<Jf~ 1 \J \ 1_ Figure 2: Two dimensional Fourier transform pulse and gradient sequence. selective 90° RF pulse together with a slice select gradient (see page 6 for description of slice selection). During data acquisition, a "read" gradient is applied, which frequency-encodes the sample under study and thus provides spatia 1 information; this functions in the same way as the gradients used in the Lauterbur method. In addition, a phase encoding gradient is applied, perpendicular to the first, and its amplitude is incremented for each successive excitation. The number of these phase encoding increments chosen determines the 2D matrix size and thus the resolution (see Chapter 1). In all images obtained for this thesis, 256 phase encoding increments were used. The observed signal S(t) can be written as S(t) - J/p(r)8(r,t)dxdy (3) where s(r,t)dxdy is the contribution from the area dxdy, volume averaged over the slice thickness, at position r, and p(r) is the spin density. The area dxdy is referred to as a pixel. The 2DFT of S(t) is given by S(w) = S(wx,a)y) (4) such that S(u>) - J/S(t)exp(-iu)t)dtxdty (5) For further details, the reader is referred to Mansfield and • 21 Morris Imaging methods may also be divided into groups depending on whether they receive signal from one point at a 22 23 24 time , from a 1ine , or from the whole sample SI ice Selection The two imaging methods described above provide spatial discrimination along the x, y-axes of the object. For an object of finite length, the resultant image represents the 6 projection of all the spin-densities onto a nominal plane. Clearly, it is necessary to have a means for selecting a slice of known thickness at any position within the object. This technique of selective excitation or slice selection was first proposed by Garroway, Grannel1 and Mansfield in 197425. Consider the irradiation of a thin cross-sectional area perpendicular to the direction of the static magnetic field, BQ (Figure 3). If a linear magnetic field gradient is applied in the Bq direction, the absorption frequency is a function of position in this direction. Irradiation with a pulse containing a narrow band of frequencies will then excite only the desired cross-section as shown in Figure 3. Narrow band of RF 7 It is desirable that the physical domain of the excited slice be rectangular in cross section. To implement this, a rectangular envelope of frequencies must be applied to the sample; in turn, this requires use of a sinc-shaped pulse in the t i me doma i n. In practice, the sine function must be truncated since it extends to infinity. The result of this is some distortion in the frequency-envelope. These distortions can be minimized by weighting the time domain with a damping function, typically a Gaussian function. The resulting RF pulse shape used for NMR imaging is shown in Figure 4, together with its Fourier transform. Figure 4: Selective radio-frequency pulse in the time domain and its Fourier transform. The response of the nuclei within this slice can be described as follows. The effective field experienced by the nuclei (Be^^) is given by equation 6. 8 Beff - Bx + GZ (6) In the absence of the gradient and assuming a sufficiently intense Bj, -the nuclear moments would precess in phase. However, due to the presence of the gradient, nuclei at different positions along BQ will experience different local fields. This results in a loss of phase coherence at the end of the 90° RF pulse. However, it can be shown for a 90° RF pulse that this variation in phase with z is approximately linear, and thus phase coherence can be recovered by z-gradient reversal for a length of time approximately half that of the irradiating pulse. The full sequence of events for slice selection are shown in Figure 2. The slice thickness is determined by the magnitude of the magnetic field gradient used, and the frequency bandwidth of the selective RF pulse. In December 1982, the prototype whole body NMR imaging system of Picker International was installed in the Extended Care Unit of the Health Sciences Centre Hospital on the University of British Columbia (UBC) campus. This system is part of the Imaging Research Centre, which also includes a CT X-ray scanner and Positron Emission Tomograph. The NMR imaging system consists of an Oxford Instruments superconducting magnet, with a room temperature bore of 1 metre, operating at a static magnetic field strength of 0.15 Tesla (proton resonance frequency, 6.4 MHz). The system is 9 interfaced with a Perk in Elmer computer, and al1 operational software is written by Picker International (see Appendix I for more details). Patient scanning for clinical research has been carried out on a volunteer basis, since NMR imaging systems have not yet been approved as diagnostic tools in Canada. Time has also been available for scientific research. The UBC instrument makes use of the imaging method of plane selection. The first software package available to the user only contained the technique of filtered back projection for obtaining an image; however subsequent packages also contained software for imaging using two-dimensional Fourier transformation. A number of different pulse sequences are available for use on the Picker International system, for example: 1. Inversion-recovery (180o_T-90°-Data acquisition-De1 ay-)n For this sequence a 180° RF pulse is used to invert the equilibrium magnetization, M along the z-axis to the -z-direction, i.e. M2 = -MQ. Spin-lattice relaxation then takes place and the magnetization along z is gradually restored to MQ (see Figure 5). Application of a 90° RF pulse at time tau (i) after the 180° pulse will then tip any magnetization along z into the x'y' plane, which will induce a signal in the receiver coll, the amplitude of which is given by equation 7. I - yi-^xpC-T/T^) (7) 10 'A i Bo b 2 A Bo 180V r c j •A :Bo s. Duringt d 3 X r BO 90V 41* Figure 5: Vector diagrams demonstrating the effects of an inversion-recovery pulse sequence on a spin-system. A plot of signal amplitude versus i gives an exponential recovery curve like that shown in Figure 6. The return of the magnetization to its equilibrium state is dependent on the spin-lattice relaxation time, T . The magnetic interactions giving rise to this relaxation are considered in more detail in Chapter 2. The actual inversion-recovery pulse sequence used on the whole body imaging system is given by equation 8. 1 1 -Mo Figure 6: Plot of intensity versus tau, the inter-pulse delay, for an inversion-recovery pulse sequence. (-180°-T-90°-t-180-t-Data aquisition-)n (8) The reasons for having the second 180° RF pulse and its implications are dealt with in Chapter 2. 2. Spin-echo (90°—t-180°—r.-Data Acqu i s i t i on-De lay-) n At this juncture it seems appropriate to introduce the 27 concept of the spin-echo . It will also be considered again in Chapter 2. The use of this pulse sequence is more clearly understood by considering the vector diagrams in Figure 7. After the 90° RF pulse (along x'in the rotating frame of reference) the magnetization MQ is turned into the x'y' plane along y'. i.e. the individual protons precess about z 12 At time 2t after 90*x-Figure 7: Spin-echo pulse sequence. Schematic diagrams showing formation of the echo. 13 in the xy plane of the laboratory frame of reference (Figure 7b). The protons then dephase relative to each other due to the magnetic field inhomogeneities and spin-spin relaxation. Some protons will precess faster than u and some slower o (Figure 7c). The spin-spin relaxation time, 1^, describes the phase memory of the spin system and hence the decay of magnetization in a particular direction in the x'y' plane after application of the 90° RF pulse. If a 180° RF pulse is applied along x' at a time x after the 90° pulse, the protons will now be rotated 180° about x' and those that were rotating clockwise in the x'y' plane will rotate counterclockwise and vice versa (Figure 7d). A further time x after the 180° pulse, the individual protons will cross the -y' axis together, and a negative signal will build up and decay in the receiver coil (this is the "echo"), after this the protons will continue to dephase. The inverting 180° pulse cancels out any static magnetic field inhomogeneity effects and in this way the amplitude of the spin-echo is dependent only on and x. The amplitude of the spin-echo for a given t-value is shown in equation 9. R(2T) = Zo exP<-2*/T2) (9) If, however, molecular diffusion takes place between the 90° and 180 pulses, then the echo amplitude will be affected by 28 the different magnetic fields experienced by individual protons as they move. The echo amplitude is then given by equat ion 10, 14 I(2T) Xo exP<-2T/T2) exP <- I Y2G2Dx3) (10) where G is the spatial magnetic field gradient and D is the diffusion coefficient. In the absence of the effects of diffusion, a plot of echo amplitude versus 2t gives an exponential plot like that shown in Figure 8. Figure 8: Plot of intensity versus 2 tau for a spin-echo pulse sequence. 3. Free Induction Decay (90°-Data acquisition-De1 ay-)n In this pulse sequence, after the 90° RF pulse the equilibrium magnetization, MQ, is turned into the x'y' plane and sampled almost immediately. The signal amplitude is given by equation 11. 1=1 (l-exp(-TR/T.)) (11) o J-15 For all three pulse sequences described above, the time between successive reapplications of the entire sequence is referred to as the repeat time (TR). The appearance of the NMR image is dependent on a number of parameters of the object; namely, spin-lattice relaxation time (Tj), spin-spin relaxation time (T2), spin-density (p), diffusion (D) and flow. It also depends on the pulse sequence used to obtain the image, which is chosen to provide an image in which the displayed intensity is weighted towards one of the object parameters. For example, the inversion-recovery pulse sequence provides an image 29 which is Tj weighted t and the contrast observed between different areas of the object is mainly due to the difference in their Tj-values. This can be explained as follows: for a given t-value, areas of different Tj will have recovered different magnitudes of z-magnetization: thus o * when the 90 sampling pulse is applied to observe the signals, areas of different T will have different intensity. As a second example, the Free Induction Decay sequence provides an image which is mainly dependent on the 30 spin density » p, or in other words, the number of protons present in any part of the sample. This is easily explained, since this sequence merely samples the equilibrium z-magnetization. However, if the delay between data acquisition and reapp1icat ion of the 90° pulse is not sufficiently long for all protons to relax back to 16 equilibrium, then the image will also display some degree of Tj dependence. Spin-spin relaxation may also occur between the 90° pulse and data acquisition, resulting in an image which also has a dependence on T_. In v i vo NMR images can now be obtained with exquisite anatomical detail (Figure 9). With this achieved, the next Figure 9: Sagittal image of a human head obtained using a spin-echo pulse sequence. goal is to quantify NMR parameters measured j_n vi vo, in the hope that they can provide a method of tissue 31 32 characterization ' . In addition, it is only with the help of quantitative measurements that there is any chance of differentiating between different types of pathology which appear visually identical on the NMR image. 17 Thesis Objectives 1. To determine whether it is possible to obtain values of the spin-lattice relaxation time, which are reproducible and have acceptable errors, under the conditions of the Imaging experiment. The effects of varying the imaging conditions on the values obtained will also be examined. 2. To determine whether it is possible to obtain values of the spin-spin relaxation time, which are reproducible and have acceptable errors, under the conditions of the imaging experiment. 3. To apply quantitative NMR imaging j_n v i vo and determine whether it is feasible to use the spin-lattice and spin-spin relaxation times to differentiate between pathology which appears visually similar on the NMR image. In addition to the usual problems of quantitative NMR measurements, the conditions of the imaging experiment introduce further possible sources of error, namely: 1. Inhomogeneity of the static magnetic field (BQ) and the RF field (Bj). In imaging, the use of large bore magnets means that field inhomogeneities are difficult to overcome. If these inhomogeneities are large, they would give rise to a range of relaxation times, since they are field dependent. 2. Magnetic field gradient non-linearity. This causes distortion in the image. 18 3. Slice selection. This introduces the problem of volume averaging which is discussed in detail in Chapter 2. 4. Spatial resolution. This will determine the smallest area which can be studied. 5. Reproducibility. If quantitative measurements are to be of diagnostic use they must be reproducible over time. 6. Methods of Reconstruction. These can affect the appearance of the image and introduce ambiguities in the observed intensities. 7. Software Limitations. Certain parameters (such as TR) have limits, which can affect the outcome of accurate quantitative measurements. At the outset, the strategy of this work was to carry out quantitative measurements on simple systems, then progress to more complex ones. This proved to be an unexpectedly difficult task. Initially, much time was expended to obtain spacially encoded xspectra' using Fourier transformation of single projections from back projection data sets (Figure 10) of vials containing water doped with various concentrations of copper(II) sulphate (CuSO^). The intention was that the signal intensities from these * spectra' could be used for calculation of TJt without artefacts from any subsequent imaging operation. It was expected that if the inversion-recovery pulse sequence was used, then, as described earlier, positive and negative signal intensities would be observed, depending on the 19 Figure 10: Fourier transform of a single projection in back projection imaging. relaxation time of the sample under study and the choice of t-value. However, when in practice a single concentration was studied using a range of i-values, the observed signal intensities were always found to be positive. This implied that the absolute sign of signal intensities around the null point was in serious doubt, and could affect the value of Tj obtained. In addition, when sets of different solutions were studied simultaneously, fluctuations in signal signs for individual solutions were observed and phase errors were apparent. An example is shown in Figure 11. In an attempt to understand these phenomena, which the manufacturers, Picker International, could not explain, various combinations of solutions doped with CuSO^ were studied. It became apparent that the phenomena arose from the methods of phasing and reconstruction employed by Picker International. These problems are considered in detail in Chapter 1. That chapter also evaluates some of the other sources of error 20 WW Figure 11: Fourier transform of a single projection in back projection imaging showing phase errors. mentioned previously, such as BQ inhomogeneity, using phantoms containing water doped with paramagnetic ions. Once the behaviour of the instrument was understood, quantitative Tj ancj measurements were carried out on these simple systems. The effect of the imaging conditions on the values obtained was studied, and complete error analyses carried out on al1 the results. The ultimate goal of the studies was, of course, quantitative applications J_n vi vo. The ability to carry out quantitative NMR measurements j_n vi vo is a difficult task. There are a number of reasons for this: 1. The mammalian cell composition and environment Is very complex and Is not static. 2. Water exists in more than one state within cells34, and if there Is slow exchange on the NMR time scale, contributions may come from each separate state giving rise 21 contributions may come from each separate state giving rise 35 to multiexponential relaxation behaviour • This means that the resultant signal intensity can be the sum of signal intensities from the various compartments having different relaxation properties. This is considered in more detail in Chapter 2. 3. Mobile fat protons may contribute to the NMR signal, again giving rise to multiexponential relaxation behaviour since the protons of fat and water typically have different relaxation rates, but virtually identical resonance frequencies at the low magnetic field strength used in i mag i ng. These obstacles present a real challenge. Chapter 3 describes the first use of quantitative NMR imaging to study the development of Experimental Allergic Encephalomyelitis 36 (EAE), a wel1 documented animal model for Multiple 37 Sclerosis (MS) , in primates. EAE was induced in a Macaca  fascicu1ar i s monkey, and the development of the disease was followed using quantitative NMR imaging. The disease gives rise to abnormal areas in the central nervous system (CNS) which are similar to those observed in humans with MS. 38 These abnormal areas appear bright on a spin-echo image-3 39 and dark on an inversion-recovery image , due to the elevation in their 1^ and Tj values respectively. Quantitative measurements were carried out on the monkey's brain from the day of detection of the disease until death. 22 mortem, and the Tj and j£ vaiues. These studies have already shown NMR imaging to be a powerful tool in the study of EAE In primates, and the future looks promising for extrapolating the findings to the study of MS in humans. 23 CHAPTER 1 EVALUATION OF THE INSTRUMENT In preparation for carrying out quantitative NMR measurements on the whole body NMR imaging instrument, it is important that its mode of operation be fully understood; the following evaluation provided the necessary Information for measuring the NMR parameters of Interest In this work. The effects of field 1nhomogeneity on the quality of images obtained using the instrument were examined. This provided a numerical value for the homogeneous volume at the centre of the magnet. Within this volume, the smallest object which could be Identified was then determined. This is referred to as the spatial resolution. The effects of automatic setting of the attenuator, and the methods of phasing and reconstruction used on the instrument were studied. The problems they Introduced were identified and explained, thereby allowing useful quantitative measurements to be carried out on this instrument. This chapter describes the process of evaluation in detail. a. Static Magnetic Field Inhomogeneity and  Rad1 ofreguency Field Inhomogeneity The usual method adopted for studying the spatial distribution of these fields involves plotting them on a 24 point by point basis, which is time consuming. It was therefore decided to construct phantoms which could provide an indication of the distribution of the static magnetic field inhomogeneity, ABq, and the radiofrequency field inhomogeneity, AB j. These phantoms consisted of equally spaced 1 cm diameter glass tubes containing water doped with CuS04, as shown in Figure 12. Using the instrument r-o U~ i 10 20 —r-30 40 1 cm Figure 12: Diagram of parallel tube phantom for studying magnetic field inhomogeneity. transmitter coil, which is also used for body imaging, 10 mm and 20 mm slice images were obtained using the Free Induction Decay sequence (FID) shown in Figure 13. Images were obtained in all three orientations, xy, xz and yz, using a 256 x 256 data matrix. An example of these images, taken in the xz plane, is shown in Figure 14. Distortions 25 90* Slice Select Gradient Read Gradient Phase Encoding Gradient Data Collect Figure 13: Free Induction Decay (FID) pulse and gradient sequence. in the images of the glass tubes represent ABQ and/or gradient non-linearity, AG. The distortions are observed on Figure 14: FID image of parallel tube phantom in the xz plane. 26 the periphery of the image which shows that, as expected, AG and ABQ become significantly larger as one moves away from the coil centre. Variation in the intensity across the image provides information on ABj. imperfect pulses will result in a variation in intensity in that region of the image. Figure 15 is a plot of intensity as a function of distance along the x-directi on; it demonstrates clearly how the intensity varies. 1—z—i—» s~ 7 Distance (cm) Figure 15: Plot of intensity as a function of distance along the x-direction. Examination of al1 these images together indicates that a volume of 12100 cm^ at the centre of the coil is homogeneous, and should provide good quality images and consistent quantitative measurements. Variations in pulse length across this region can be corrected for in the data processing. Outside this region the homogeneity 27 deteriorates, and although qualitatively useful images can be obtained out to 45 cm in diameter and of 45 cm along the magnet axis, intensity and spatial distortions are both preva1ent. b. Spat ia 1 Resolut ion A simple phantom was constructed to determine the spatial resolution of the instrument used. It consisted of 180' 180* RF J IBCf 7 1 8*o» Select Gradient Read Gradient Pnase CncodinQ Gradient Data Collect J L Figure 16: Inversion-recovery pulse and gradient sequence. melting-point capillary tubes filled with two different solutions of CuS04? rj.99 mM and 2.54 mM, randomly arranged in vials of 2 cm diameter. 10 mm horizontal slice Images were then obtained using the inversion-recovery sequence given In Figure 16. As can be seen In the Image In Figure 28 17, it appears that individual capillary tubes can be identified which suggests that 1 mm resolution is I 1 ' 1 « 1 • 1 ' 1 1 0 2 4 6 8 cm Figure 17: Inversion-recovery image of capillary tube phantom. attainable. However, it is possible that the signal intensity observed in a single pixel is arising from more than one tube. A second phantom was therefore constructed, and is shown in Figure 18; all dimensions are as indicated. Figure 18: Diagram of resolution phantom. 29 In the image shown in Figure 19, the arrow indicates the resolution of the 1 mm perspex wall between two of the XL Figure 19: Spin-echo image of resolution phantom. areas. This demonstrates that 1 mm is indeed easily resolvable and the maximum resolution possible is somewhat less than this. c. Attenuation The instrument used in these studies is designed such that it automatically changes the input gain of the RF receiver in order to maximize the signal to be digitized. This implies that for a series of inversion-recovery images with a range of -t-values (where t = pulse interval), the 30 attenuation will not be constant and the Intensities not directly comparable. The instrument, however, also allows manual setting of the attenuator. For all studies in this thesis, the attenuation was set based on the signal from the Figure 20: Phantom used for studying attenuation and for carrying out Tj and T2 measurements. sample using a spin-echo sequence with 2t = 26 ms and TR = 5 sec. In order to check whether manual setting of the attenuation provides the same intensities as automatic setting, a simple experiment was carried out. The phantom shown in Figure 20, which contained three solutions of CuSO^, was placed in the receiver coil and two images obtained using the spin-echo sequence with Zi = 26 ms and TR = 5 sec. The first image was obtained with automatic setting of the attenuator, which gave 26 decibels (dB). The 31 second image was then obtained by manually setting the attenuation to 26 dB. The intensities obtained for each solution are given in Table I. Table I: Intensities from two spin-echo images comparing manual and automatic setting of the attenuation. Intensities are given as the mean values from 2.6 cm2 plus or minus standard deviation, at the centre of the vials. Two values are given for each solution corresponding to two different positions in the receiver coil. Attenuation 4.96 in 2.54 all 0.99 if) Autoaatic Manual 123 t 7.6 133 i 6.1 122 i 7.6 131 t 5.9 140 i 6.5 134 i 6.5 139 • 6.6 133 J 6.5 154 i 7.4 146 i 9.7 152 • 7.6 145 i 7.9 The intensities observed in both images are virtually identical. It can therefore be assumed that manual setting of the attenuation will provide images with the correct i ntens it f es. d. Phas ing and Reconstruct ion As<^previous 1 y considered, the contrast observed In images of an object which has regions of different Tj values, obtained using the inversion-recovery (IR) pulse sequence 12, is.dependent on the intrinsic difference 32 (-180°-T-900-Data acquisition -) (12) n between the spin-lattice relaxation rates of the protons within the different spatial domains. This leads to intensity changes which are a function of the pulse interval, i. A typical situation for a two compartment sample is shown in Figure 21 for two solutions of CuSO. 4 (see Chapter 2). The signal intensity observed from a single compartment at a particular value of T is given by 40 equation 13: Ix - Io{l-[l-rW(l-exp(-k/T1)]exp(-T/T1)} (13) where IQ = intensity at T >> Tj k = wa i t i ng per i od between data acqu i s i t i on and reapplication of the 180° pulse W = correction factor for 180° pulse. Although the NMR data-acquisition Itself produces signals which have the phase information necessary to distinguish between the positive and negative-going signals associated with the magnetization plot shown in Figure 21a, many imaging protocols can only produce NMR signals in the "absolute mode". What this means is that after acquisition of the time domain signal and subsequent Fourier transformation, all signals are rectified to become positive. In those circumstances, the form of the 33 inversion-recovery intensity curve corresponds to Figure magnitude reconstruction. 21b, and the *high intensity' signals observed in a reconstructed image could equally well come from a region of long Tj as from one of short Tj. In order to minimize the occurrence of this ambiguity, the software of the instrument used in this study has the option of "rea1-intensity" reconstruction, which provides either positive or negative numbers, as appropriate, when an IR pulse sequence is employed. However, during attempts to obtain values of Tj from various systems, it was observed that this method of reconstruction introduced additional problems. In order to Illustrate the phenomenon observed, two simple phantoms were constructed, each consisting of a 20 mm Figure 22: Phantom used for studying intensity ambiguities A = 4.96 mM CuS04, B = 0.99 mM CuS04 diameter vial located inside a beaker of 70 mm diameter (Figure 22). The two aqueous solutions contained CuS04.5H20 at accurately known concentrations of 4.96 mM and 0.99 mM and having known Tj values of 117 ms and 534 ms respectively (see p.60). As assembled, phantom-A corresponds to a * hot-spot' and phantom-B to a * cold-spot' in terms of relative Tj va1ues. The two phantoms were placed side by side In the receiver coil used for head imaging, which has an aperture of 30 cm, and horizontal (xz) 10 mm thick slice-images were obtained simultaneously through both. IR images were 35 obtained with t-values ranging from 50-1000 ms and TR of 4 sec plus T. The resultant images are shown in Figure 23. The positions of the two phantoms were then reversed and the experiments repeated; as expected, the relative signal magnitudes of each particular solution were unchanged. The white horizontal line through each image shows the position of the cursor, and the trace below it, the corresponding intensity-projection. The intensity trace shows clearly the positive and negative-going signals, as can be seen in Figure 23c where the outer-component of the right-hand phantom and the inner component of the left-hand phantom are both negative. 36 Figure 23: Inversion-recovery images demonstrating intensity ambiguities. a. i = 50 ms c. X = 150 ms e. X = 300 ms g. 1 = 500 ms i . X = 1000 ms b. x — 10 0 ms d. x — 250 ms f. x = 400 ms h. x - 600 ms 37 Since phase-sensitive reconstruction was employed to obtain the IR-images in Figure 23, one would expect the relative intensities of each of the two solutions to follow the trend shown in Figure 21a; however, as can be seen both from the cursor traces themselves and the plot of intensity versus T shown in Figure 24, this is not the Figure 24: Plot of intensity versus tau showing fluctuations in signal sign. • 4.99 mM CuSC- © 0.99 mM CuSO. 4 4 case. The nature of the problem is more clearly understood when one examines the data corresponding to i = 50 ms, at which time both signal intensities should be negative, whereas they are, in fact, positive. As a second example, given that the null point of the 4.96 mM CuS04 solution is close to i = 100 ms, all signals observed after that i-time should be positive. In practice the value at t = 150 ms is 38 negative and remains so until T = 250 ms at which time it becomes positive. Conversely, the signal intensity of the 0.99 mM solution should be negative prior to its null point at x = 400 ms, and thereafter become positive. Instead, the signal is positive from i = 50-150 ms, negative between 250-400 ms, after which time it becomes positive again. Obviously, these fluctuations in signal sign made any attempt at quantitative measurements very difficult, since the absolute sign of the signal in a particular image was in serious doubt. However, this phenomenon can be explained41 and compensated for. The centre view of a 2D image has effectively no phase encoding gradient. The instrument software carries out a peak search on this view to set the attenuation based on the magnitude of the maximum intensity peak. A global phase correction Is then carried out by making this maximum peak real and positive. All other signal signs are then relative to this peak. If it happens to be a negative-going peak then it Is phase-corrected 180° to become the maximum positive peak. Even when the attenuation is manually set, this phase correction Is still carried out, resulting In the same signal-sign fluctuations. Thus, when i = 50 ms, the signals from both the 0.99 mM and 4.96 mM solutions are negative, with the maximum intensity signal coming from the 0.99 mM solution. This automatically acts as the reference peak and is inverted 180°. Since the 39 signal from the 4.96 mM solution Is of the same sign, it too becomes positive. As another example, consider the Image obtained when T = 150 ms. At this t-value the signal from the 0.99 mM solution should be negative, and that from the 4.96 mM solution positive; in practice the reverse is observed. This is due to the fact that the magnitude of the signal from the 0.99 mM solution is still larger than that from the 4.96 mM solution, by virtue of Its longer TJt and hence it again acts as the reference peak. In order to get around the problem of fluctuating signal signs, the solutions under study were surrounded by a fast relaxing water bath (~10 mM CuS04 solution). This bath provided the most signal and was always positive for the range of i-values being used. It therefore always acted as the reference peak and consequently ensured the correct phasing of al1 other signals. One unfortunate result of this, however, was that it was no longer possible to use the single projections from back projection images for measuring Tj, since they were too complex. Instead IR images were obtained, the intensities for each concentration of paramagnetic species measured from them directly, and subsequently processed. In most cases, images obtained j_n vivo will not be affected by this method of phasing. However, if a region of long Tj has a large volume, such as a fluid-filled cyst, then the negative signal which should be observed from this 40 region on an Inversion-recovery image is. In fact, positive. The reason for this is the signal is phase corrected 180° to become the maximum positive peak. The result of this, of course, is that the signals from the surrounding regions are also inverted and have the wrong sign. It is clearly a problem which the manufacturer will have to deal with. One method of phasing employed by other companies is to use a paired saturation-recovery image as a reference. Although this method removes the possibility of intensity ambiguities, it does have the disadvantage of having a longer total imaging time. A further instrumental aberration was observed and is shown in the t = 200 ms IR-image in Figure 25a, which consists of light and dark bands across the entire phantoms. The measurement was repeated on a large number of occasions over a period of months with Identical results, ruling out the possibility of a transient instrumental problem. At this t-value, the signal from the 4.96 mM sample should be positive and that from the 0.99 mM negative. That these two signals have approximately equal magnitudes is shown in the image in Figure 25b as absolute intensities following magnitude reconstruction. The banding can be explained41 as follows: when the instrument carries out its peak search on the centre view of the 20 image, it is unable to find a maximum peak. The Instrument effectively sees no signal due to the intensities 41 having equal but opposite sign. This results in a shift in the echo position, that is, there is a time shift of the time domain signal. This can be represented in the Figure 25: a. Inversion-recovery image of phantom used to demonstrate intensity ambiguities at i = 200 ms. b. The same image in a. after magnitude reconstruct i on. 42 following way: if f(t) has the Fourier transform F(u)» then the function f(t-a) has the transform F(w)e~1ua 42. The derivation of this is straightforward: f(t-a)e-i(Jtdt - /f(t-a)e-iw(t-a)e-iuad(t-a) - e^FCa.) (14) In other words, if the time domain signal is shifted in time, its Fourier transform remains identical except that it gets multiplied by a phase factor which varies linearly with the frequency, u. and is proportional to the time shift, a. This results in the phase shifts observed in the image in Figure 25a. Additional proof that the banding observed is due to a time shift of the time domain signal was provided by the following experiment. As indicated previously, the actual IR pulse sequence used on this instrument has an additional 180° pulse which is applied at time t after the 90° pulse and subsequent data acquisition of the echo at 2t. In order to collect the echo at this time, the correct gradient sequence must be used. If t = 13 ms, then the read and phase encoding gradients required for the data acquisition are identical to those used for a spin-echo sequence with 2T = 26 ms. If the time of data acquisition is left unchanged, and the above gradients substituted with those required for a spin-echo sequence that has 2t = 20 ms, the result is a shift of the echo position in the time domain signal. This substitution gave rise to banded images like that shown in 43 Figure 25a. It is clear from these studies that any factor which causes a shift In time of the time domain signal, will give rise to the phase shifts observed on these images. It was also possible to correct for these phase shifts using the fast relaxing background water bath, since there was no longer any possibility of the instrument misinterpreting the signals from the solutions being studied. This allowed useful data to be obtained for the range of T-values required to determine T,. 44 CHAPTER 2 QUANTITATION OF NMR PARAMETERS The appearance of an NMR Image Is dependent on more than one parameter. Pulse sequences are chosen to provide images whose intensities are weighted towards one parameter, and which also provide the necessary differences In intensity between normal and diseased tissue for identification of pathology. The contrast provided by Tj and T2~weighted images are the most important. The optimum choice of pulse sequence is the one which provides the most tissue discrimination; this in turn depends on the relative Tj'and T2_values. This chapter describes the measurement of Tj and T2 using the Picker International whole body imaging instrument. Both time constants are defined and the methods of measurement are described. The effects of imaging conditions and instrument limitations on the values of Tj and T^ obtained are examined, a. Spin-Lattice Relaxation (Tp (i) Definition The spin-lattice relaxation time (Tj) can be defined as the time constant of the process whereby an ensemble of nuclear spins return to thermal equilibrium with their lattice after perturbation. This form of relaxation involves an exchange of energy between the spin and the lattice and occurs due to Interaction between the 45 fluctuating field generated by the precessing nuclei and those generated in the lattice by movement of other magnetic nuclei. These interactions can involve a number of different processes43, namely: 1. magnetic dipole-dipole interactions 2. electric quadrupole interactions 3. chemical shift anisotropy interactions 4. scalar-coupling interactions 5. spin rotation interactions. Any mechanism which gives rise to fluctuating magnetic fields at a nucleus is a possible relaxation mechanism. All aqueous solutions studied in this section contained paramagnetic species. Since the magnetic moments of the unpaired electrons in these paramagnetic species are £a. 103 times greater44 than the nuclear magnetic moments of the water protons, they give rise to much larger local fields, and hence dominate the relaxation of the water. In a dilute solution of a paramagnetic species, the relaxation of the solvent nuclei (In this case water) will be dominated by the effects of the unpaired electrons on the solvation sphere of the metal ion45. The bound water molecules and the bulk solvent have different relaxation times. Rapid exchange of the water molecules and the protons of the molecules within the two phases gives rise to a single observed relaxation time. If the random fluctuating magnetic fields at the site 46 of the paramagnetic species occur at the Larmor frequency of the solvent protons then T. relaxation can occur, i.e. it The first terms arise from the dipole-dipole interaction between the electron-spin, S, and the nuclear spin, I, which is characterized by the correlation time, i The second terms arise from modulation of the scalar interaction (often called isotropic nuclear-electron spin-exchange interaction) which is characterized by correlation time, t The electronic and nuclear Larmor precession frequencies are given by oi^ and wj , yj is the gyromagnetic ratio, 8 is the Bohr magneton, S is the total electron spin, r is the distance between the nucleus and the paramagnetic ion, and A/fi is the electron-nucleus hyperfine coupling constant in Hz. The correlation times are defined as follows: induces a nuclear spin transition from I = -1/2 to +1/2. The Tj of a nucleus bound to a paramagnetic site is given by the following equation : + (16) 47 where T.^ = life-time of a nucleus in the bound site tR = rotational correlation time of the bound paramagnetic ion ig = electron-spin relaxation time. (ii) Measurement of Tj The utilization of Tj in NMR spectroscopy, including methods of measurement, is well documented^-5*} # In this study, the method known as inversion-recovery was used to carry out the measurements on the aqueous solutions, and is described below. The pulse sequence used in the inversion-recovery method is given below: (-180°-T-90°-Data acquisition -) (12) The intensity at a particular t-value is given by equation 7, I = Io(l-2exp(-T/T1)) (7) which is obtained by integration of the Bioch equation describing decay of Mz, dM —5- - -(M -M )/T. (18) dt N z o 1 where Mq = thermal equilibrium value of M , 48 The intensity is measured over a range of t-values, then Tj may be obtained from an exponential fit of the data or a semi-log plot. The repetition time must be sufficiently long (5Tj) to allow the nuclei to return to their equilibrium state before reapplication of the 180° pulse. In the imaging experiment the z-component of the observable magnetization may not be fully inverted due to the radiofrequency field inhomogeneity over the large volumes being studied. Incomplete inversion of the magnetization can be corrected for in the data processing by fitting the data to equation 13, which includes a correction factor, W, for the 180° pulse40, IT - Io{l-[l-W(l-exp(-k/T1)]exp(-x/Tl)} (13) where I = intensity at x >> T, o 1 k = waiting period between data acquisition and reapplication of 180° pulse W = correction factor for 180° pulse. A phantom consisting of 2 cm diameter vials placed in a large petri dish was constructed as shown in Figure 20 (p.31). The vials were filled with accurately known concentrations of CuSO^ and manganese(I I) chloride (MnC^) solutions, and Tj measurements were carried out using the inversion-recovery method. 49 Copper(II) Sulphate Solutions Inversion-recovery horizontal (xz) slice images were obtained with t-values in the range 50-1000 ms and a TR of" 4 sec plus x. The mean intensities over 2.6 cm2 at the centre of the vials were obtained directly from the images using the cursor on the instrument and a three-parameter exponential fit was carried out on the data using equation 13. Manganese(11) Chloride Solutions Inversion-recovery images were obtained as before with x-values in the range 10-1000 ms and TR of 3 sec plus t. The intensities were obtained once again from the centres of the vials, and processed in the same way as the CuS04 data. A 957. accuracy linear regression analysis was carried out on all results, and the Tj values, plus or minus two standard deviations, are given in Table II. 50 Table II: Tj values for various concentrations of CuS04 and MnCl2 solutions obtained at 20 (± 1)°C using the inversion-recovery method. Two values are given for each concentration corresponding to two different positions in the receiver coil. So 1ut f on Concentration (mM) Tj (ms) CuSC-4 4.96 108 ± 10 4.96 126 ± 4 2.54 245 ± 6 2.54 243 ± 13 0.99 519 ± 4 0.99 548 ± 2 MnCl 2 0.67 103 ± 1 0.67 106 ± 1 0.34 189 ± 2 0.34 188 ± 1 0. 16 350 ± 8 0.16 340 ± 8 0.06 659 ± 9 0.06 663 ± 8 51 (iii) Computed Tj The software of the instrument, developed by Picker International, has the capability of carrying out Tj computat ions. Two images are required to carry out these Tj computations: an inversion-recovery and a spin-echo image. The repeat times must be equal in both images. The intensities for each image are given in equations 19 and 20. IK - iMrt-ltflJll-d-Zexpi-VTJ + expU-k-O/TpiexpC-t/T^ (19) I - Ioexp(-2x/T2)[l-2exp(-k/T1) + exp^TR/T^ ] (20) IIR l~(2-2 exp(-k/T1) + exp(-x-k/T1)exp(-t/T1) (n) ISE (l-2exp(-k/T1) + exp(-TR/T1) The ratio of these intensities is given in equation 21. Taking the ratio eliminates the image intensity dependence on the equilibrium magnetization and T2, then Tj may be obtained from a look-up table, which is a plot of SQ versus Tj. TJ computations were carried out on solutions containing the same paramagnetic Ion concentration as in (ii) and the numbers compared. The results are given in Table III. The computed T2 values are given as the mean value from 2.6 cm2 at the centre of the vials. 52 Table III: Computed Tj values for various concentrations of CuS04 and MnCl2 solutions obtained at 20 (± 1)°C Two values are given for each concentration, corresponding to two different positions in the receiver coi1. So1ut i on Concentration (mM) Computed Tj (ms) CuSC-4 4.96 1 17 4.96 123 2.54 245 2.54 235 0.99 556 0.99 548 MnCl 2 0.67 83 0.67 98 0.34 192 0.34 188 0. 16 357 0. 16 350 0. 06 678 0.06 679 53 (iv) Effects of a Spin-echo Readout After application of the 90° RF pulse in an IR sequence, not all the nuclear spins will precess at the same Larmor frequency; they will be out of phase. This is due to the presence of the field gradient. Unless the spins are in phase, the xy magnetization is not detected. In order to get around this, a second 180° refocusing pulse is applied after a short time (13 ms) and the spin-echo collected 2t (26 ms) after the 90° pulse. This is referred to as a spin-echo readout. During the time 2t, spin-spin relaxation can occur, which may introduce a further source of error in the measurement of T}, since the equation for the intensity at any t-value will have not only the exponential Tj term but also an exponential T2 term. The method of data processing used in (ii) does not take into account the fact that a SE readout is employed. In theory, it should be possible to treat the exponential T2 term as a constant, since it is identical in every sequence. In order to check this, two different spin-echo readouts were used and the Tj values compared. Spin-echo t-values of 13 ms and 20 ms were employed. The data for t = 13 ms is given in Table II. The use of a SE readout will have more of an effect on short T2 values; therefore this experiment was only carried out on the MnCl2 solutions, where T2 is considerably shorter than T,. The results for a t-value of 20 ms are shown in Table 54 IV, together with the computed Tj values using the same readout. Table IV: Tj values for various concentrations of MnCl2 solution, obtained at 20 (± i)°C using the inversion-recovery method, and a spin-echo readout with T = 20 ms. Computed Tj values with the same readout are given for comparison. Two values are given for each concentration, corresponding to two different positions in the recei ver coi1. Solution Concentration (mM) Tl (ms) Computed Tj (ms) MnCl 2 0.67 100 + 3 78 0.67 107 + 10 73 0.34 191 7 190 0.34 189 ± 5 189 0. 16 343 + 3 346 0. 16 335 3 352 (v) Effects of Proton Density The signal intensity observed on an NMR image is dependent not only on the relaxation times of the protons but also on the number of protons giving rise to the signal. This is referred to as the proton density. A number of slice thicknesses are available for use on the imaging 55 instrument. Varying the slice thickness may introduce a further source of error in the measurement of Tj, since it changes the proton density contribution to the observed signal . If, for a series of IR images, the same slice thickness is applied for each image, it should be possible to treat the proton density factor as a constant. This was tested by using two additional slice thicknesses of 5 mm and 20 mm and running inversion-recovery Tj experiments on the MnCl2 solutions. The images were processed as before and the results are shown in Tables Va and Vb, together with computed Tj values with the same slice thickness. Table Va: Tj values for various concentrations of MnCl2 solution, obtained at 20 (± i)°C using the inversion-recovery method, and a slice thickness of 20 mm. Computed T^ values are given for compar i son. So1ut1 on Concentration (mM) Tl (ms) Computed Tj (ms) MnCl 2 0.34 198 + 2 174 0.34 182 + 5 178 0. 16 330 + 1 1 349 0. 16 350 6 338 0.06 630 13 682 0.06 689 22 675 56 Table Vb: values for various concentrations of MnCl2 solution, obtained at 20 (± i)°C using the inversion-recovery method, and a slice thickness of 5 mm. Computed Tj values are given for compar i son. So 1ut i on Concentration (mM) Tl (ms) Computed Tj (ms) MnCl2 0.34 183 4 187 0.34 186 ± 3 184 0. 16 349 + 2 333 0. 16 346 + 1 337 0.06 662 + 12 650 0.06 696 4 683 (vi) Multiexponential Relaxation Behaviour Mu1tiexponential relaxation behaviour may be observed for an isotropic solution containing protons with different physical characteristics, such as a mixture of oil and water, if proton exchange does not occur or is slow on the NMR time scale. In an imaging measurement, additional sources give rise to mu11iexponentia 1 relaxation behaviour; of these, volume averaging Is the most significant. An image can be obtained by selectively exciting a slice 10 mm thick. Within this thickness there may be water protons in different compartments or in different types of tissue. 57 which are not exchanging. The signal contribution from each proton source depends on its volume and on the pulse sequence used to excite the slice. The resulting image intensity in each pixel is the sum of the signal contributions through the slice. For proton NMR imaging studies of mammalian tissues, the signals of water and fat will overlap due to the negligible chemical shift difference between the two signals at the low magnetic field used to obtain the images. This could also give rise to multiexponential relaxation behaviour, since the mobile fat protons may have a different Tj and T2 than those of water. If more than one exponential decay is present, the signal intensity observed for a particular t-value is given by equation 22. I(T) - I ^i*1-2 EXP(-^T1±)) <22> where n = number of exponentials required for an adequate description of the data, taking into account experimental accuracy, . Computer programs are available for analysis of relaxation data. These programs fit a chosen number of exponentials to the data, and thus provide the different relaxation times contributing to the signal. Consider two compartments, between which there is no exchange, and having Tj values of 245 ms and 519 ms. If they are excited simultaneously in the same slice, then the 58 observed signal intensity for a given t-value in an inversion-recovery image is given by the sum of the intensities for each compartment. The resultant magnetization plot, together with the individual plots for each compartment, are given in Figure 26, and the observed Tj is 338 ms. This means that the computed Tj images can Figure 26: Plot of Intensity versus tau demonstrating multiexponential relaxation behaviour. O 2.54 mM CuS04 A 0.99 mM CuS04 X 2.54 mM CuS04 +0.99 CuS04 only provide observed Tj-values, giving no indication as to their source. In order to determine whether the observed T is the result of multiexponentiaI relaxation behaviour, the intensity must be measured as a function of t. 59 (vii) D i scuss i on The Tj data are summarized in Figure 27. The results show that useful Tj measurements can be made in the range 100-600 ms. Although this is not a large range, it does encompass the Tj-values of many tissues, and in particular those of the grey and white matter of the brain. Comparison Figure 27: a. Plot of spin-lattice relaxation rate versus concentration for CuSO^ solutions. O IP data A Computed data b. Plot of spin-lattice relaxation rate versus concentration for MnCl2 solutions. O IR data & Computed data 60 of the Tj data obtained using the inversion-recovery method with that from the computed images gives a correlation within 107.. The values obtained from the computed Images are therefore useful, since they can be obtained more rapidly than the Inversion-recovery data. However, it is not possible to make error estimates on these numbers, or to extract information on multiexponential relaxation behaviour. Taking into account the random error In the data together with the effects of inhomogeneity across the coil, reproducible values of Tj can be obtained ± 12%. In most cases the errors were somewhat less than this. The same trend was observed for data obtained using a longer spin-echo read time and again when the slice thickness was varied. Comparison of the mean values and errors obtained for data under different imaging conditions shows a correlation of less than 57. between the results. The data demonstrates that over the Tj range studied, it is possible to treat many conditions of the imaging experiment as constants without affecting the measured Tj-values. The Tj values are within 127. of literature values measured under conventional spectroscopic conditions, with no magnetic 51 field gradients present . These findings contradict those reported in the literature by Rosen, Pykett and Brady52, who state that corrections must be made to T. rjata for the effects of slice 61 selection and a spin-echo readout. Multiexponential relaxation behaviour arising from volume averaging is a problem in imaging. It is important to be aware of it when applying NMR Imaging J_n vi vo. The observed T^ —Va1ues may arise from more than one tissue type; however, the values can be compared if the same slice is excited in serial studies. This is considered again in Chapter 3. b. Spin-spin Relaxation (T2) (1) Definition Immediately after a spin system has been perturbed by a 90° pulse, for example, the spins of the individual nuclei are in phase. They then begin to exchange energy with one another leading to the decay of the magnetization in the x'y' plane, M ,, with a time constant T2. In addition, inhomogeneities in the static magnetic field, AB0, will result in a range of Larmor precession frequencies which also cause M , to decay. Overall My- decreases to zero with a time constant T2* which includes a term for the field inhomogeneity (equation 23). i i **o — - — + (23) * T 2 T2 2 Fluctuating fields In the x, y and z directions all have an effect on T2„ The T2 of a nucleus bound to a paramagnetic site is given by the following equation 24.4^ 62 1 2 2 2 1 Y\& S(S+1)B c + 13x c (24) 15 6 r 1+U)TX 2 2 l+0)oT 2 2 + ls(S+l) -3 h x e + x e The parameters are the equivalent of those in the Tj equation (p.47) and the terms arise from the same types of interactions as those described for Tj. (i i) Measurement of As previously mentioned, the magnetization decay is dependent not only on T2, but on the static magnetic field inhomogeneity. It is therefore not possible to use T2 as a measure of T^. 27 In 1950, Hahn proposed the use of the spin-echo method to overcome the dephasing effect of the field inhomogeneity. The principle behind the spin-echo technique was shown in Figure 6, page 13. Immediately after application of the 90° pulse, all the nuclear spins are in phase, i.e.'they are all precess.ing with the same Larmor frequency. The presence of any inhomogeneity in the static magnetic field means that not all nuclei will be experiencing the same external field, and the result of this is dephasing of the spins. If the spins are allowed to dephase for a time x, followed by application of a 180° pulse about the x-axis, they will refocus or come back into 63 phase at time 2t, when they can be detected. The intensity is measured as a function of Zx, thus providing a value of by carrying out either a logarithmic plot or an exponential fit. The repeat time for the sequence must be sufficiently long to allow return to equilibrium between successive pulses, just as in the IR Tj experiment. The use of the spin-echo method is limited however, since molecular diffusion may occur during the refocusing of the magnetization resulting in a reduction in the echo 28 amplitude. Carr and Puree 11 have shown that the effect of diffusion on a spin-echo experiment is dependent on the spatial magnetic field gradients (G), the diffusion coefficient (D) and the time during which diffusion can occur. It is given by equation 10. I(2T) " Io exP(-2*/T2) exp (- | Y2G2D-c3) (10) 3 The x dependence means that the effects of diffusion will be much more pronounced for large values of T^. The effects of diffusion can be overcome, to a great extent, by using 29 the Cari—Puree 11 method , which Is a modification of the Hahn spin-echo method. The Cari—Puree 11 pulse sequence is given in equation 25, (90°-T-180°—c-[echo]-T-180°-x-[echo]- ) (25) Positive and negative echoes are alternately formed. This method has two advantages: it saves time and by making x 64 short the effects of diffusion may be virtually eliminated. One drawback is that pulse imperfections may cause incomplete rephasing of the spins resulting in error. This sequence is not available on the instrument used for these studies. The same phantom constructed for the Tj measurements was also used for the T2 measurements. Horizontal (xz) SE slice images were obtained, with i-values in the range 13-160 ms and TR of 3 sec plus t, for the CuS04 and MnCl2 solutions. The intensities were again obtained directly from the images and an exponential fit to equation 9 was carried out to determine T2, X(2T) " Xo -PC-^'V (9> where I = equilibrium magnetization. A 957. accuracy linear regression analysis was carried out on al1 T2 values obtained, and values plus or minus two standard deviations are given in Table VI. 65 Table VI: T2 values for various concentrations of CuS04 and MnCl2 solution, obtained at 20 (± 1)°C using the spin-echo method. Two values are given for each concentration corresponding to two different positions in the receiver coil. Solution Concentration (mM) T2 (ms) CuSO. 4 4.96 1 16 ± 4 4.96 116 ± 4 2.54 194 ± 10 2.54 191 ± 15 0.99 313 ± 22 0.99 354 ± 26 MnCl2 0.67 46 ± 2 0.67 46 ± 3 0.34 78 ±3 0.34 78 ± 5 0.16 136 ± 7 0. 16 136 ± 4 0.06 244 ± 14 0.06 248 ± 9 66 (i i i) Computed Two spin-echo images are required to carry out a 1^ computation using the Picker International Instrument software. These images must have equal repeat times and different echo times. The ratio of these intensities is then taken, thus eliminating the image intensity dependence on the equilibrium magnetization and Tj. J2 is then obtained from a look-up table, just as for 7 . The ratio is given in equation 26. T2 computations were carried out on the same solutions as in (fi), using spin-echo t-times of 20 ms and 40 ms, and the results are given in Table VII. ISE. (l-ZexpC-k./T.) + exp(-TR/T ) exp(2x /T ) S i = — - (26) ISE (l-2exp(-k2/T1) + exp(-TR/T1) exp(2T2/T2) 67 Table VII: Computed T2 values for various concentrations of CuS04 and MnCl2 solutions, obtained at 20 (± 1)°C. Two values are given for each concentration corresponding to two different positions in the receiver coll. Solution Concentration (mM) T2 (ms) CuSC* 4 4.96 120 4.96 131 2.54 219 2.54 221 0.99 522 0.99 548 MnCl2 0.67 43 0.67 44 0.34 77 0.34 78 0. 16 135 0. 16 137 0.06 414 0.06 443 68 The effect of varying the combination of t-time on the resultant T2 values was also examined. Computed T2 measurements were carried out on the MnCl^ solutions, using different i-times, and the results are given in Table VIII. Table VIII: Computed T2 values for various concentrations of MnCl2 solutions, obtained at 20 (± 1)°C using various combinations of i-times. Two values are given for each concentration corresponding to two different positions in the receiver coil. Concentration (mli) 20,40 t-times (ms) 20,60 40,60 0.67 43 43 44 0.67 44 43 46 0.34 77 77 71 0.34 78 71 69 0. 16 135 124 123 0. 16 137 129 122 (iv) Effects of Diffusion Only the SE method is available for measuring T2 using the whole body imaging instrument. As expected, when attempts were made to measure long T2 values, the effects of diffusion resulted in abnormally 69 low values of T^. A maximum value of 200 ms could be measured. The effects of diffusion on the computed T2 values are minimized because a ratio Is taken and, as a result, above T2 = 200 ms the computed values are higher than those obtained using the exponential fits to the SE data. (v) Discussion The T2 data is summarized in Figure 28. For T2 values in the range 40-200 ms, the difference between spin-echo data and the computed T2 values is 15% or less. The computed T2 images are useful, since not only can they provide a quick Indication of the T2_va]ue t,ut they also minimize the effects of diffusion, thereby allowing the study of longer T2. However, just as for the computed Tj values, it is not possible to make error estimates or to extract information on multiexponential relaxation behaviour from these numbers. Above the range of 40-200 ms, the effects of diffusion Increase, as one would expect. The TJ/T2 ratio for water doped with CuS04 jS close to unity, and once the T2 value gets above 200 ms, the spin-echo data does not reflect this. Comparison of literature values51 obtained at the same field using the Carr-Purcel1 method, as modified by Meiboom and Gill53, shows a correlation of a few percent for the range 40-200 ms. 70 Although this is a narrow range, many tissue T2~values lie within it, in particular those of brain tissue. i " 3 ' 5 * I oS oTi oTT^ Concent rat ion (mM) Concent ration (mM) Figure 28: a. Plot of spin-spin rate versus concentration for CuSO. 4 solutions. O SE data & Computed data b. Plot of spin-spin relaxation rate versus concentration for MnC^ solutions O SE data A Computed data 71 CHAPTER 3 Applications of Quantitative NMR Imaging This chapter describes In detail an experimental application of quantitative NMR imaging j_n v i vo. The reader is Introduced to the structure of the brain and spinal cord, and the diseases of interest in this study. How these diseases affect the central nervous system is also exp1 a i ned. (i) Background The nerve cells, or neurons of the brain and spinal cord consist df cell bodies and axons, as shown in Figure 29. It is along the axon that the nerve impulses travel, and the axons of different neurons communicate with one dendrites axon Figure 29: Diagram of nerve cell. another at junctions known as synapses. Many axons are surrounded by a myelin sheath, which Increases the rate at 72 which the axon can conduct impulses. The myelin allows what is known as saltatory conduction. This is where conduction is by a mixture of cable properties of the nerve fibres, as well as by chemical mechanisms. In demyelinating diseases, such as Multiple Sclerosis (MS), this myelin sheath is broken down, resulting in a decrease in the speed at which a nerve impulse can travel along the axon. The areas of abnormality in MS are large (> 3 mm), and of more than one type. They can occur anywhere In the central nervous system, predominantly in the white matter but also in the grey matter. One of the major problems in mapping the course of MS Is that the only data available on humans Is post mortem, by which time the disease Is usually in the chronic stage. This means that little is known about the pathology of the early stages of the disease. For this reason, animal models for MS have been developed in order to study the mechanism of the development of the early lesion. One such model is Experimental Allergic Encephalomyelitis (EAE). NMR imaging has provided an opportunity for serially studying both MS and EAE with no known risk to the subject. The ability to detect abnormal areas is a result of changes in the NMR characteristics of the protons. The molecular environment of the protons changes, which in turn affects their NMR properties. It is these molecular changes that must eventually be charted, since they are responsible for 73 the changes observed as gross pathology and on the NMR images. This was the basis for undertaking these studies of EAE in primates. It is known that the first event in EAE is inflammation leading to haemorrhagic necrosis. Eventually demye1ination is also present. In this project serial studies have shown that the progression of EAE can be followed using NMR imaging. Furthermore, measurement of Tj and T^ as a function of time provides an indication of change on a molecular level ±n v i vo. (ii) Induction of EAE and NMR Imaging Protocol EAE was induced in a male Macaca fasc i cu1ar i s monkey, weighing 3 kg, by injection of 0.15 mL of a water-in-oil emulsion containing 15 mg monkey myelin basic protein and 0.5 mg heat killed Mycobacter i um tubercu1os i s intradermally 54 in the hindfoot pads . The animal was obtained from the Bureau of Biologies Breeding Colony in Ottawa and was housed first in the UBC Animal Care Facility on South Campus Road, then after induction of the disease, he was moved to the Acute Care Unit animal care facility. NMR imaging data were collected on the Picker International whole body imaging system using a receiver coil with an aperture of 15 cm. Data were collected using multi-slice spin-echo and inversion-recovery pulse sequences, both of which provided 8 contiguous 5 mm thick 74 slices. Echo delays (2t) of 40 ms and 60 ms were employed in the spin-echo sequences. A i-time of 400 ms was used for the inversion-recovery sequence, and in all sequences the repeat time was 2 sec. These parameters were chosen in order to allow direct comparison with human MS data, and the choice of pulse sequences would allow computation of Tj and V As indicated in Chapter 2, volume averaging can give rise to multiexponential relaxation behaviour. When images are obtained from the monkey's brain, both grey and white matter may be excited in the same slice. If the results of serial scanning were to be comparable, then accurate repositioning of the monkey's head was essential; the same slices had to be excited in each set of images obtained. This was achieved In the following way: using Polyflex-11, a commercially available thermoplastic polymer used In the PET program, a mould was made of the monkey's head, and fitted to the receiver coll. This allowed repositioning within 2 mm for each set of scans. Figure 30 shows the position of the monkey in the scanner and the slice images being obta i ned. 75 Figure 30: Diagram showing the position of the monkey in the instrument and the slices being obtained. (iii) Development of EAE Using NMR Imaging The monkey was anaesthetized for imaging using 0.8 mL combination of ketamine and rompun, (12:1 ratio) the effects of which lasted 1.5-2 hours. He was checked three times every 24 hours for the onset of clinical signs^5 and scanned daily until the detection of the first abnormal area, after which time he was scanned approximately every 10 hours. The monkey developed a definite abnormal area in the white matter of the left cerebral hemisphere, 16 days after inoculation and before the onset of any obvious clinical signs. The image obtained on that day Is shown in Figure 31. A spin-echo pulse sequence with an echo-delay of 40 ms was used to obtain the image. The arrow indicates the abnormal area. After the 180° refocusing pulse is applied 76 In the spin-echo sequence, the signal intensity observed from the abnormal area is greater than that from normal white matter due to an increase tn T_f a decrease in its relaxation rate. The abnormal area therefore appears brighter than normal tissue on a spin-echo Image. Figure 31: Transverse SE image of the monkey's brain, showing the abnormal area in the left hemisphere (right side of image). The development of the disease could be easily followed. Figure 32 shows a series of spin-echo images obtained from the monkey. The number of days after inoculation are as Indicated in the caption. 77 Figure 32: Series of three spin-echo Images obtained from the monkey's brain. a. 16.25 days after inoculation b. 16.75 days after Inoculation c. 18.42 days after Inoculation The light areas are abnormal. 78 (iv) Quantitation of NMR Parameters Tj and computations were carried out on the abnormal areas from the day of appearance until death. Table IX and X contain sample series of these measurements, obtained from the same slice during serial scanning. Table IX: T} values as a function of time after induction of EAE, for white matter (WM) and grey matter (GM) which appear normal, and the lesion. Time (Days) WM Tj (ms) GM Lesion 16.25 370 530 440 16.75 370 530 480 17.12 390 520 500 17.67 380 530 540 18. 04 390 530 570 18.42 370 510 590 18.92 370 520 600 79 Table X: T^ values as a function of time after induction of EAE, for white matter (WM) and grey matter (GM) which appear normal and the lesion. Time (Days) WM T2 (ms) GM Lesion 16.25 110 110 150 16.75 110 110 150 17.12 120 140 200 17.67 120 1 10 180 18.04 120 100 220 18.42 100 90 240 18.92 1 10 110 240 Table XI contains Tj data for different slices, obtained from the monkey's brain immediately prior to death. These values, together with those for 1^, were correlated with the pathology found post mortem. 80 Table XI: Tj values from five different slice images immediately prior to death for white matter (WM) and grey matter (GM) which appear normal, and the lesion. SI ice WM Tj (ms) GM Les ion 1 390 490 610 2 390 450 600 3 390 470 650 4 390 470 560 5 400 500 660 (v) Discussion These studies were based on the postulate that NMR imaging can be used to detect and follow the development of EAE in primates. Experimental results have 56 shown this to be true EAE can be detected in primates before the onset of clinical signs using NMR imaging. This has implications for the study of MS in humans; some MS lesions appear unaccompanied by new clinical symptoms. It may be that the first lesion observed in the monkey, which was not reflected clinically, is pathologically similar to clinically asymptomatic MS lesions in humans. Characterization of these asymptomatic lesions in MS 81 could provide the information necessary to understand and chart the progression of the disease. This is an exciting topic for further exploration. The progress of EAE in primates can be easily followed using NMR imaging. The time of appearance of individual lesions could be noted, which a 1 lowed mapping of the disease post mortem. Until now, this information on the whole disease process has been unobtainable. Accurate repositioning is essential, since volume averaging could lead to discrepancies in the appearance of the image, and in quantitative measurements, if more than one tissue type is contributing to the signal observed from a particular slice. Comparison of the relaxation data obtained immediately prior to death with the histopathology, reveals that longer Tj and T2 values are associated with the presence of inflammation, haemorrhagic necrosis and demyelination. In the initial stages of the disease, not all three types of pathology are present. The changes in Tj and T2 over time reflect the molecular changes occurring due to the progression of the disease. The individual changes in Tj and T2 from detection until death (607. and 1407. respectively) are sizeable compared with the errors of 127. and 157. found in the studies on water doped with paramagnetic species, described In Chapter 2. This means that the initial lesion can be distinguished from those occurring later by these changes in T. and T9. In other 82 words, ft should be possible to distinguish between areas of inflammation (which is an early event) and areas containing demye1ination (which occurs later in the progress of the disease.) In addition to this, there is an elevation in Tj before the lesion is visible on the NMR image. If this elevation is reproducible, and is large in comparison to the errors involved, then this indicates that a fresh lesion can be detected by changes in Tj before it is visible on the image. The results of this work have shown that quantitative NMR imaging has the potential for answering many pertinent questions regarding EAE in primates and eventually MS In humans. 83 CONCLUSIONS It has been shown that reproducible values of the spin-lattice relaxation time (Tj), in the range 100-600 ms, can be obtained using the inversion-recovery method, at a field strength of 0.15 Tesla on water doped with various concentrations of paramagnetic species. It has also been shown that changing the image slice select gradient and the spin-echo read time do not affect the resultant Tj values. Using scatter in results observed under a range of conditions, it has been estimated that the uncertainty in the Tj values obtained using the inversion-recovery method is less than 127.. This method produces Tj values which are consistent with the results of the two-point computational method, and in addition, makes error estimates possible. The inversion-recovery method can also provide information on mu1tiexponentia1 relaxation behaviour, which is Important for j_n v i vo studies. The Tj values obtained are within 127. of literature values measured under conventional spectroscopic conditions with no magnetic field gradients present. Results show that reproducible values of the spin-spin relaxation time (T2) in the range 40-200 ms, can be obtained using the spin-echo method, at a field strength of 0.15 Tesla on water doped with paramagnetic species. Above this range the effects of diffusion 84 become Important, resulting in abnormally low values of T2, compared with literature values obtained using the Carr-Purcel1 method, as modified by Meiboom and Gill. The errors estimated for T2 fn the range 40-200 ms are ± 157. and the results are within 127. of the literature values. The spin-echo method provides T2 values in the range 40-200 ms, which are consistent with values obtained using the two-point computational method, and in addition makes error estimates possible. The computational method takes the ratio of two spin-echo images with different T,-values, which minimizes the effects of diffusion and provides realistic values for T2 longer than 200 ms. NMR imaging can detect Experimental Allergic Encephalomyelitis (EAE) in primates before the onset of clinical signs. The technique can be used to follow the development of the disease, which allows mapping of its pathological progression. The progression of the disease is accompanied by a 607. increase in Tj, and a 1407. Increase 1 n T2. The results indicate that the percentage changes in Tj and T2 taken together can be used to discriminate between areas of inflammation and others which contain demyel1 nation. 85 FUTURE WORK NMR and EAE in primates a) Studies of multiexponential relaxation behaviour in  v i vo. b) J_n vitro Tj and T2 measurements on normal and abnormal brain tissue. Identification of mu1tiexponentia1 relaxation behaviour of water and fat protons at high field. c) J_n vitro NMR spectroscopy on molecules other than water and fat. d) High field imaging and spectroscopy j_n vi vo i) 1H i mag i ng i i) 31P i mag i ng iii) *H spectroscopy iv) 23Na imaging e) Immunological studies on the monkey to correlate them with quantitative measurements, and compare with human data. NMR and MS in Humans a) Work has already begun on post mortem NMR studies of brain tissue before and after fixation. Correlations have been made between the NMR images and the gross pathology5^. Correlations are now being made between the Tj ancj T2 values observed post mortem and after fixation, with the histopathology. The results of 86 these studies can then be compared with those from the monkey; the ultimate goal being the identification of different types of MS pathology in humans on the NMR image using NMR parameters. b) Quantitation of NMR parameters jjn vi vo and comparison with primate data. 87 REFERENCES 1. Puree 1) E.M., Torrey H.C. and Pound R.V. Phys. Rev. 69, 37 (1946). 2. Bloch E., Hansen W.W. and Packard M. Phys. Rev. 70, 474 (1946). 3. Williams D.H. Chem. Soc. Rev., 13(2), 131 (1984). 4. Damadian R. Science 171, 1151 (1971). 5. Moon R.G. and Richards J.H. J. Biol. Chem. 248, 7276 (1973). 6. Henderson T.O., Costello A.J.R. and Omachi A. Proc. Nat. Acad. Sci. USA 71., 2487 (1974). 7. Hoult D.I., Busby S.J.W., Gadian D.G., Radda G.K., Richards R.F. and See ley P.J. Nature (London) 252, 285 ( 1974) . 8. Dawson M.J., Gadian D.G. and Wilkie D.R. Nature (London) 274, 861 (1978). 9. Nunnally R.L. and Bottomley P.A. Science 211, 177 (1980). 10. Gordon R.E., Han ley P.E., Shaw D., Gadian D.G., Radda G.K., Styles P., Bore P.J. and Chan L. Nature 287, 736 (1980). 11. Alger J.R., Sellerud L.O., Behar K.L., Gillies R.J., Shu 1 man R.G., Gordon R.E., Shaw D. and Han ley P.E. Science 214, 660 (1981). 12. Radda G.K., Bore P.J., Gadian D.G., Ross B.D., Styles 88 P., Taylor D.J. and Morgan-Hughes J. Nature 295, 608 (1982) . 13. Ross B.D., Radda G.K., Gadian D.G., Rocker G., Esiri M. and Falconer-Smith J. N. Engl. J. Med. 304, 1338 (1981). 14. Brenton D.P., Garrod P.J., Krywawich S. , Reynolds E.O.R., Bachelard H.S., Cox D.W. and Morris P.G. Lancet I, 115 (1985). 15. Farrar T.C. and Becker E.D. Pulse and Fourier Transform NMR: Introduction to Theory and Methods. New York, Academic Press (1971). 16. Lauterbur P.C. Nature 242, 190 (1973). 17. Bracewel1 R. The Fourier Transform and Its Applications. New York, McGraw-Hill (1965). 18. Brooks R.A. and Di Chiro G. Phys. Med. Biol. 2J_, 689 (1976). 19. Kumar A., Welti D. and Ernst R.R. J. Magn. Reson. 18, 69 (1975). 20. Aue W.P., Bartholdi E. and Ernst R.R. J. Chem. Phys. 64. 2229 (1976). 21. Mansfield P.A. and Morris P.G. NMR Imaging in Biomedicine. Ed. Waugh J.S. London: Academic Press ( 1982) . 22. Hinshaw W.S. J. Appl. Phys. 47, 3709 (1976). 23. Hinshaw W.S., Bottomley P.A. and Holland G.N. Nature (London) 270, 722 (1977). 89 24. Sutherland R.J. and Hutchinson J.M.S. J. Phys. E: Sci. Instr. JLL. 79 (1978). 25. Garroway A.N., Grannell P.K. and Mansfield P. J. Phys. C. 7, L457 (1974). 26. Taylor D.G. and Inambar R. In Physical Principles and Clinical Applications of Nuclear Magnetic Resonance. Ed.; Lerski R.A. Paradigm Print, Gateshead, p.23 (1985). 27. Hahn E.L. Phys. Rev. 80, 580 (1950). 28. Carr H.Y. and Puree 11 E.M. Phys. Rev. 94, 630 (1954). 29. Simmonds D., Banks L.M., Steiner R.E. and Young I.R. Neuroradio1ogy 25, 113 (1983). 30. Young I.R., Bailes D.R., Burl M., Collins A.G., Smith D. T., McDonnell M.J., Orr J.S., Banks L.M., Bydder G.M., Greenspan R.H. and Steiner R.E. J. Comput. Assist. Tomogr. 6, 1 (1982). 31. Herfkens R., Davis P., Crooks L., Kaufman L., Price D., Miller T., Margulis A.R., Watts J., Hoenninger J., Arakawa M. and McRee R. Radiology 141. 211 (1981). 32. Fullerton G.D., Potter J.L. and Dornbluth N.C. Magn. Reson. Imag. 1, 209 (1982). 33. Crooks L.E., Arakawa M., Hoenninger J., McCarten B., Watts J. and Kaufman L. Radiology 151, 127 (1984). 34. Cope F.W. Biophys. J. 9, 303 (1969). 35. Bakker C.J.G. and Vriend J. Phys. Med. Biol. 29(5), 509 (1984). 90 36. Raine C.S. Lab. Invest. 50.(6), 608 (1984). 37. Matthews W.B., Acheson E.D., Batchelor J.R. and Weller R.O. McAlpines' Multiple Sclerosis. London, Churchill Livingstone, Incorporated. (1985). 38. Bailes D.R.* Young I.R., Thomas D.J., Straughan K., Bydder G.M. and Steiner R.E. Clin. Radiol. 33, 395 (1982). 39. Young I.R., Hall A.S., Pal lis C.A., Bydder G.M., Legg N.J. and Steiner R.E. Lancet II, 1063 (1981). 40. Levy G.C. and Peat I.R. J. Magn. Reson. J_8, 500 (1975). 41. Stewart W.A. and Hall L.D. Phys. Med. Biol. In press. 42. Fukushima E. and Roeder S. Experimental Pulse NMR: A Nuts and Bolts Approach. Addison-Wes1ey, Massachusetts, p.48 (1981). 43. Abragam A. The Principles of Nuclear Magnetism. Eds.; Marshall W.C. and Wilkinson D.H. Oxford University Press (1961). 44. Dwek R.A. Nuclear Magnetic Resonance (NMR) in Biochemistry: Applications to Enzyme Systems. Eds.; Harrington W. and Peacocke R. Oxford University Press, p.175 (1973). 45. Ibid. p.177. 46. Lynch L.J. and Webster D.S. J. Magn. Reson. 40, 259 (1980). 47. Freeman R. and Hill L.D.W. J. Chem. Phys. 54(8), 3367 91 (1971). 48. McDonald G.G. and Leigh J.S. Jr. J. Magn. Reson. 9, 358 (1973). 49. Hanssum H. J. Magn. Reson. 45, 461 (1981). 50. Sass M. and Ziessow D. J. Magn. Reson. 25., 263 (1977). 51. Pykett I.L., Rosen B.R., Buonanno F.S. and Brady T.J. Phys. Med. Biol. 28(6), 723 (1983). 52. Rosen B.R., Pykett I.L. and Brady T.J. J. Comput. Assist. Tomogr. 8(2), 195 (1984). 53. Meiboom S. and Gill D. Rev. Sci. Instrum. 29, 688 (1958). 54. Alvord E.C., Shaw C. and Hruby S. Ann. Neurol. 6, 469 (1979). 55. Alvord E.C., Shaw CM., Hruby S., Sires L.R. and SI imp J.C. In Experimental Allergic Encephalomyelitis: A Useful Model for Multiple Sclerosis. Eds.; Alvord E.C., Kies M.W. and Suckling A.J. New York, Alan R. Liss, Incorporated. p.461 (1984). 56. Stewart W.A., Alvord E.C., Hruby S., Hall L.D. and Paty D.W. Lancet. II, 898 (1985). 57. Stewart W.A., Berry K., Hall L.D. and Paty D.W. Lancet II, 412 (1984). 92 APPENDIX I Block Diagram of Picker International Imagi ng Instrument ii 93 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0059464/manifest

Comment

Related Items