Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

SPECT/CT quantification of ¹⁷⁷Lu for dosimetry in radionuclide therapy treatments of neuroendocrine tumors Uribe Muñoz, Carlos Felipe 2016

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

Item Metadata

Download

Media
24-ubc_2016_may_uribemunoz_carlos.pdf [ 7.09MB ]
Metadata
JSON: 24-1.0223889.json
JSON-LD: 24-1.0223889-ld.json
RDF/XML (Pretty): 24-1.0223889-rdf.xml
RDF/JSON: 24-1.0223889-rdf.json
Turtle: 24-1.0223889-turtle.txt
N-Triples: 24-1.0223889-rdf-ntriples.txt
Original Record: 24-1.0223889-source.json
Full Text
24-1.0223889-fulltext.txt
Citation
24-1.0223889.ris

Full Text

SPECT/CT Quantification of 177Lufor Dosimetry in RadionuclideTherapy Treatments ofNeuroendocrine TumorsbyCarlos Felipe Uribe MuñozB.Sc. Physics, Universidad de los Andes, Bogota, Colombia,2008BA. Sc. Mechanical Engineering, Universidad de los Andes, Bogota,Colombia,2009M.Sc. Medical Physics, The University of British Columbia, Vancouver, Canada,2012.A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinThe Faculty of Graduate and Postdoctoral Studies(Physics)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)February 2016© Carlos Felipe Uribe Muñoz 2016AbstractPeptide receptor radionuclide therapy (PRRT) shows promising results in the treat-ment of neuroendocrine tumors. These tumors over-express somatostatin recep-tors that allow us to label somatostatin analogues with 177Lu to deliver dose tothe tumor. However, currently every patient receives the same amount of radioac-tivity of approximately 7400 MBq per treatment cycle. With this “one dose fitsall” approach, differences between patients are not taken into account resulting insome being under-treated while others over-treated. The aim of this thesis was todevelop a simple protocol for 177Lu activity quantification for patients undergo-ing PRRT with the purpose of performing personalized dose assessments. Physicsphenomena that influence image quantification were investigated.As electrons emitted in the decay of 177Lu result in creation of Bremsstrahlung,Monte-Carlo simulations were performed to investigate this effect on image quan-tification of 177Lu. Phantom experiments with different attenuation and scatterconditions were performed to test quantification accuracy and evaluate perfor-mance of several segmentation methods. Images were reconstructed using theOSEM algorithm and two scatter correction methods were compared. An exper-iment to measure the camera deadtime was performed by adding activity into aiiAbstractbottle placed in a cylindrical phantom. Plots for observed count rate vs. truecount rate were made, and the deadtime was calculated based on the paralyzablemodel. The protocol was applied to four patient data, and OLINDA and voxelizeddosimetry calculations were used to create dose volume histograms for the kid-neys. Lastly, a graphical user interface to allow for the quantitative reconstructionof the data obtained using any of the manufacturers was developed.Our results suggest that Bremsstrahlung contributions to the detected energyspectrum in imaging studies of 177Lu have no degrading effects in image quantifi-cation. Our protocol recovers the activity in kidneys to within 10%. The deadtimecorrection based on paralyzable model was accurate for the count rates measured.The deadtime corrections should be performed based on scatter corrected pho-topeak window instead of the full spectrum. Lastly, the dose delivered to thekidneys in patient data was lower than the suggested dose per session in order toreach current toxicity limits.iiiPrefaceThe material presented in Chapter 3 has been submitted for publication. Uribe C.,Esquinas P., Gonzalez, M., Celler A. Characteristics of Bremsstrahlung Emissionsof 177Lu, 188Re, and 90Y for SPECT/CT Quantification in Radionuclide Therapy.I performed the simulations, the experiments, and conducted the analysis with thehelp from Pedro Esquinas, and wrote the manuscript. The project was done withthe guidance of Dr. Anna Celler. Dr. Marjorie Gonzalez helped in experimentalstudies. Preliminary results regarding the Bremsstrahlung analysis were presentedat the IEEE NSS/MIC Conference in October 2013, Seoul, Korea.The material presented in Chapter 4 has been submitted for publication. UribeC., Esquinas P., Tanguay J., Odette G. E. Gonzalez M., Beauregard JM., CellerA. Accuracy of Quantification of 177Lu SPECT Images: A Phantom Study. Iperformed the experiments with the coauthors help. Data received from Que-bec was provided by Emilie Odette Gaudin and Dr. Jean-Mathieu Beauregard. Iperformed the analysis with the guidance and discussion with the coauthors andwrote the manuscript. Preliminary results were presented at the Annual Congressof EANM in Gothenburg, Sweden, October 2014, the IEEE NSS/MIC Confer-ence in November 2014, Seattle, USA, and the Annual Congress of EANM inivPrefaceHamburg, Germany, October 2015.The material presented in Chapter 5 is being prepared for submission. UribeC., Tanguay J., Gonzalez M., Esquinas P., Celler A. Deadtime Effects in Quantifi-cation of 177Lu for Radionuclide Therapy. I performed the experiments with thehelp of the coauthors, performed the analysis, and wrote the manuscript. Prelimi-nary results regarding this topic were presented at the Annual Congress of EANMin Gothenburg, Sweden, October 2014.The material presented in Chapter 6 was presented at the the Annual Congressof EANM in Gothenburg, Sweden, October 2014.The material presented in Chapter 7 is being prepared for submission. UribeC., Celler A. SPEQToR: A quantitative SPECT/CT reconstruction tool. I incorpo-rated functions developed by the Medical Imaging Group at UBC into a softwarepackage, and coded the graphical user interface. It allows for the implementationof the developed 177Lu protocol presented in this thesis. All this was done withthe guidance of Dr. Anna Celler.The Ethics Committee of CHU de Quebec approved the patient scans. Becauseof the retrospective nature of the analysis, the need to acquire patient consent waswaived.vTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiiList of Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xviiiAcknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxii1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Neuroendocrine Tumor Diagnosis . . . . . . . . . . . . . . . . . 21.2 NETs Treatment . . . . . . . . . . . . . . . . . . . . . . . . . . 51.3 Outline of the Thesis . . . . . . . . . . . . . . . . . . . . . . . . 122 Background of Nuclear Medicine and Dosimetry . . . . . . . . . . 152.1 What is Nuclear Medicine? . . . . . . . . . . . . . . . . . . . . 15viTable of Contents2.2 Nuclear Decay . . . . . . . . . . . . . . . . . . . . . . . . . . . 172.3 Interaction of Radiation with Matter . . . . . . . . . . . . . . . 252.4 Nuclear Medicine Imaging . . . . . . . . . . . . . . . . . . . . 322.5 Dosimetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 382.6 Closing Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . 483 Bremsstrahlung Characteristics of 177Lu . . . . . . . . . . . . . . . 493.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493.2 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . 523.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 603.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 663.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 714 Quantification of 177Lu in Phantoms . . . . . . . . . . . . . . . . . 734.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 734.2 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . 784.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 894.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 964.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1015 Deadtime Corrections . . . . . . . . . . . . . . . . . . . . . . . . . 1035.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1035.2 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . 1045.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110viiTable of Contents5.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1165.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1206 Dosimetry in Patients Undergoing PRRT for NETs Treatment . . . 1226.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1226.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1276.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1306.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1356.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1377 Single Photon Emission Quantitative Tomographic Reconstruction (SPE-QToR) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1387.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1387.2 Loading the Data . . . . . . . . . . . . . . . . . . . . . . . . . . 1397.3 Data Display . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1427.4 Specifying the Reconstruction Protocol . . . . . . . . . . . . . . 1447.5 Running the Reconstruction and Determining the NormalizationFactor of the Camera . . . . . . . . . . . . . . . . . . . . . . . . 1477.6 Testing the GUI and Quantification . . . . . . . . . . . . . . . . 1497.7 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1507.8 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1508 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1528.1 Summary and Findings . . . . . . . . . . . . . . . . . . . . . . . 152viiiTable of Contents8.2 Contribution to the Field . . . . . . . . . . . . . . . . . . . . . . 1568.3 Suggestions for Future Work . . . . . . . . . . . . . . . . . . . . 156Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158ixList of Tables1.1 Probability of detection of pancreatic Neuroendocrine tumors (NETs)with different imaging modalities. [1] . . . . . . . . . . . . . . . 32.1 List of nuclear families with some examples. . . . . . . . . . . . 202.2 Decay properties for 90Y, 188Re, and 177Lu (only gamma emis-sions with abundance higher than 1% are shown). . . . . . . . . . 253.1 Characteristics of the SymbiaT collimators which were used in thesimulations and experimental acquisitions discussed in this work(data from Siemens [2]). . . . . . . . . . . . . . . . . . . . . . . 563.2 Mean energy of Bremsstrahlung photons (mean Bremsstrahlungenergy - mean Bremsstrahlung energy (MBE)) created by mono-energetic electrons and β decay of 177Lu. . . . . . . . . . . . . . 614.1 Energy window setup used for phantom experiments of 177Lu . . . 814.2 Phantoms used in planar and tomographic acquisitions in deter-mination of the camera normalization factor for both 99mTc and177Lu. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 844.3 Summary of the different phantom experiments. . . . . . . . . . 90xList of Tables5.1 Energy window limits for the two photopeaks of 177Lu. Addi-tional energy windows were used to collect data for the wholespectrum. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1055.2 Coefficient of determination (R2) of the fits of the data to the par-alyzable model. . . . . . . . . . . . . . . . . . . . . . . . . . . . 1125.3 Window fractions (in percentage) for the two photopeaks, detec-tors, and SPECT cameras. . . . . . . . . . . . . . . . . . . . . . 1155.4 Values for τw obtained with equation 5.4 using values of η avail-able in literature. Window fractions obtained from detector 2were used as it covered a larger range of count rates. The lasttwo columns show the value of η required to correctly obtain thedeatime (DT) value of our experiments. . . . . . . . . . . . . . . 1156.1 Dose to the kidneys for four patients undergoing PRRT estimatedusing three different methods. . . . . . . . . . . . . . . . . . . . 1356.2 Normalized dose per injected activity for the kidneys of four pa-tients undergoing PRRT. . . . . . . . . . . . . . . . . . . . . . . 1357.1 List of collimator specifications currently available in the GUI.More collimators can be easily added by editing a file text and in-putting the manufacturer, type of collimator, hole diameter, length,and septal thickness of the collimator. . . . . . . . . . . . . . . . 144xiList of Figures1.1 Biochemical markers used for diagnosis of NETs. (Copyright El-sevier, taken from Modlin et al. [1] with permission.) . . . . . . . 32.1 Decay scheme of 177Lu into 177H f . . . . . . . . . . . . . . . . . 242.2 Diagram of photoelectric effect. An incident gamma is absorbedby the atom and is used to eject an electron. . . . . . . . . . . . . 272.3 Diagram of the Compton effect. An incident gamma interacts witha weakly bound electron and scatters. The scattered photon hasenergy lower than the incident photon. . . . . . . . . . . . . . . 292.4 Mass attenuation coefficients for water, lead, and NaI as a functionof photon energy. (The data to generate these plots was obtainedfrom the National Institute of Standards and Technology web-page http://physics.nist.gov/PhysRefData/Xcom/html/xcom1-t.htmlon November 2015). . . . . . . . . . . . . . . . . . . . . . . . . 31xiiList of Figures2.5 Picture of a Siemens SymbiaT SPECT/CT camera system. Thiscamera has two detectors attached to a gantry. They can rotatearound the patient in order to collect projections at different an-gles. An x-ray tube is also present allowing for the collection ofanatomical data using CT. . . . . . . . . . . . . . . . . . . . . . 322.6 Flow diagram of the OSEM algorithm. An initial estimate of theimage is provided and updated on each iteration until the desiredimage is obtained. . . . . . . . . . . . . . . . . . . . . . . . . . 362.7 Activity washout within a patient. The effective half life has to bedetermined in order to be able to find the total number of emittedparticles in order to perform dosimetry calculations. . . . . . . . 402.8 Diagram of attenuation of photons within the object being scanned.Attenuated photons do not reach the detector and this translatesinto an underestimation of activity if this effect of attenuation isnot corrected for. . . . . . . . . . . . . . . . . . . . . . . . . . . 412.9 Diagram of the scatter of photons that degrade image quality. Pho-tons can scatter within the object and be detected giving the im-pression that they are originate at a different location from wherethe source is located, and it also contributes to a higher number ofcounts which is translated into higher activity values. . . . . . . . 432.10 A typical detected spectrum of 177Lu is shown in a dashed line.The spectrum is composed by the primary photons (red) and scat-tered photons (blue). . . . . . . . . . . . . . . . . . . . . . . . . 44xiiiList of Figures2.11 Behavior of camera count rate detection as a function of the truecount rate reaching the detector. The ideal situation in which theobserved count rate is the same as the true count rate is shown inblue. Deviations from this line due to the paralyzable (black) andnon-paralyzable (red) models are also shown. The plots have beenmade assuming a value of τ = 5µs. . . . . . . . . . . . . . . . . 473.2 Comparison of total Bremsstrahlung yield (TBY) and Bremsstrahlungyield for photons with energies higher than 50keV (BY50) ob-tained from simulations using GATE and MCNP5. The yields ofBremsstrahlung photons generated by mono-energetic electrons(a) and by the β emissions from the radioisotopes (b) are shown. . 613.1 Bremsstrahlung spectra generated by (a) mono-energetic electronsand by (b) the β emissions from 177Lu generated with GATE(solid lines) and MCNP5 (dashed lines). The spectra were nor-malized to the total number of primary electrons. . . . . . . . . . 623.3 Total yields of Bremsstrahlung (Total Bremsstrahlung Yield (TBY))and non-Bremsstrahlung (non-BRS yield (NBY)) photons gener-ated by β ’s emitted by 177Lu, in the water-filled cylinder and thecorresponding total yields of photons that escaped the phantom. . 633.4 Analysis of energy spectra of photons recorded by the SiemensSymbiaT SPECT camera and those obtained from GATE simula-tions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64xivList of Figures3.5 Percentage contributions of each type of photons to the total num-ber of detected photons for different collimators. . . . . . . . . . 654.1 Work flow for the generation of quantitative 177Lu images. Imagereconstruction with attenuation and scatter correction, in combi-nation with a correctly determined camera normalization factorgenerate an image containing the activity distribution. Segmenta-tion for the determination of the dose in region of interest (ROI)’sis performed once the activity distribution has been determined. . 794.2 Camera calibration factors obtained with the two planar acquisi-tion methods and the tomographic scans for 99mTc (a) and 177Lu (b).. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 914.3 Quantification accuracy for the different phantom inserts scannedin air (a) and its distribution (b) for both scatter correction meth-ods, TEW and APDI. . . . . . . . . . . . . . . . . . . . . . . . . 934.4 Quantification accuracy for the different phantom inserts scannedin water (a) and its distribution (b) for both scatter correctionmethods, TEW and APDI. . . . . . . . . . . . . . . . . . . . . . 944.5 Quantification accuracy for the different phantom inserts scannedin hot water and segmented with three different methods: 40%fixed threshold (a), CT based (b), and IADT (c). The distributionof accuracy is shown in (d) for both scatter correction algorithms,TEW and APDI. . . . . . . . . . . . . . . . . . . . . . . . . . . 95xvList of Figures5.1 Coronal view of the positioning of the phantom and the bottleinside it with respect to the two detectors. . . . . . . . . . . . . . 1065.2 Measured spectra recorded at different count rates for both detec-tors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1115.3 Observed count rates as a function of true count rates. The valuesof τ from the paralyzable model are shown for both detectors andboth cameras. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1135.4 Deadtime correction factors for observed count rates in a SiemensSymbia and GE Hawkeye SPECT/CT cameras. . . . . . . . . . . 1146.1 Images of the four patients scanned on injection day t0, one dayafter injection t1, and three days after injection t3. . . . . . . . . . 1296.2 Patient 1 TAC and DVH. . . . . . . . . . . . . . . . . . . . . . . 1316.3 Patient 2 TAC and DVH. . . . . . . . . . . . . . . . . . . . . . . 1326.4 Patient 3 TAC and DVH. . . . . . . . . . . . . . . . . . . . . . . 1336.5 Patient 4 TAC and DVH. . . . . . . . . . . . . . . . . . . . . . . 1347.1 “Load Data” section of the GUI. This image shows a Nuclearmedicine and µ-map loaded from a Siemens camera. (The cameramodel is shown as specified in the DICOM file header. . . . . . . 1407.2 Example header of a nuclear medicine data file. . . . . . . . . . . 1417.3 Nuclear Medicine projections displayed on the left, attenuationmap shown in the middle, and the optional CT slices at the right. . 142xviList of Figures7.4 Quality control graphical user interface (GUI) to investigate thelinogram and sinogram of the data. . . . . . . . . . . . . . . . . 1437.5 Section of single photon emission quantitative tomographic re-construction (SPEQToR) to define the reconstruction parameters.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1457.6 Example of a list of saved protocols with the details for the firstone at the bottom. . . . . . . . . . . . . . . . . . . . . . . . . . 1487.7 Input information to determine the sensitivity or normalizationfactor for the camera. In this case, an example of a sensitivityscan with 1000MBq of 177Lu, performed at 12:15 on December18 of 2014 is being shown. For this case in particular, a sensitivityfactor of 0.6Counts/(min∗ kBq) was obtained. . . . . . . . . . . . 1497.8 Full view of SPEQToR. . . . . . . . . . . . . . . . . . . . . . . . 150xviiList of AcronymsAC attenuation correctionAPDI analytical photon distribution interpolatedBB broad beamBq BecquerelBRS BremsstrahlungBY50 BRS yield for photons with energy higher than 50 keVCF calibration factorCi CurieCT computed tomographyDT deatimeDTCF deadtime correction factorsDVH dose-volume histogramEC electron capturexviiiList of AcronymsENSDF evaluated nuclear structure data fileFBP filtered back projectionFOV field of viewFWHM full width half maximumGATE Geant4 application for tomographic emissionGEANT4 Generation and tracking V.4GUI graphical user interfaceHE High EnergyIADT iterative adaptive dual thresholdingLEHR Low Energy High ResolutionLK left kidneyLSW low scatter windowMBE mean Bremsstrahlung energyMC Monte-CarloMCNP5 Monte Carlo N-Particle V.5ME medium energyMEGP medium energy general purposexixList of AcronymsMELP Medium EnergyMIRD Committee on Medical Internal Radiation DoseMIRG Medical Imaging Research GroupMLEM maximum likelihood expectation maximizationMRI magnetic resonance imagingNBY non-BRS yieldNETs Neuroendocrine tumorsNM Nuclear medicineOLINDA organ level internal dose assessmentOSEM ordered subsets expectation maximizationPET positron emission tomographyPMT photomultiplier tubesPRRT peptide receptor radionuclide therapyPVE partial volume effectPW photopeak windowRADAR RAdiation Dose Assessment ResourceRK right kidneyxxList of AcronymsROI region of interestRR resolution recoveryRTOG radiation therapy oncology groupSBR signal to background ratioSC scatter correctionSEER surveillance, epidemiology, and end resultsSPECT single photon emission computed tomographySPEQToR single photon emission quantitative tomographic reconstructionSSR somatostatin receptorsSST somatostatinTAC time activity curveTBY Total Bremsstrahlung YieldTEW triple energy windowTRT Targeted radionuclide therapyUSW upper scatter windowVOI volume of interestxxiAcknowledgmentsI would like to thank my supervisor Dr. Anna Celler for letting me join the Med-ical Imaging Research Group back in early 2012. Anna, thanks for the supportand patience, for pushing me to the limits in order to do things the best pos-sible way. Thanks for giving me so many opportunities to attend conferences,meet colleagues, and travel around the world. You did not only provide me witha great academic experience someone could have ever dreamed about during aPhD, but you also showed me the importance of having a good work-life balancewith friends, family, and adventures. Thanks for the hard work and the numerousdiscussions in order to make this project a reality. I am grateful for all the supportduring these last 4 years. You are definitely a role model to follow.I also want to thank the members of my PhD committee: Dr. Cheryl Duzenli,Dr. Francois Benard, and Dr. Stefan Reinsberg. Thank you for all the feedbackand discussions.Thanks to the staff from the Nuclear Medicine Department at the VancouverGeneral Hospital. In particular, Dr. Marjorie Gonzalez, thank you for all the helpwith my experiments, all the time spent dealing with the cameras, the good jokesand laughs, and for being not just a colleague but also a good friend.xxiiAcknowledgmentsDr. Jean-Mathieu Beauregard and his master’s student Emilie Gaudin, ourcollaborators from Quebec City, thank you for providing us with patient data andfor all the time and discussions that made this project better.To all the patients enrolled in this research project, although I never got tomeet you, I understand the difficult times you were going through. Thank you fortrusting us, and you can be sure that we make the best efforts to improve and saveyour lives and the lives of future generations.Thanks to the University of British Columbia and the National Sciences andEngineering Council of Canada for funding my research.Thanks to all my colleagues and friends from the MIRG group for all the helpwith my project, for the lunches and dinners, and for the fun that we had out ofthe lab. You all managed to make every working day an enjoyable experience.Special thanks to Pedro for all the help with the Bremsstrahlung simulations.Mom and Dad, this has finally become a reality, I am about to become thefirst doctor in the family. Thank you so much for all the love you have given methroughout my whole life. Thank you for all the great moments we have had asa beautiful family and for all the sacrifices you have done for Caro and me. Atthe end, everything has proven that we can achieve big things with dedication andeffort. All that you have done for us so far has been worth it. I hope we can haveyou for many more years to come, do know that I am very proud of you as parentsand I do think I have had the best ones I could possibly have. This PhD degree isfrom you and for you.Caro, my little sister, thank you for being the best. Thanks for being there inxxiiiAcknowledgmentsall the good and bad moments, thanks for all the help with things that I neededto do back home but could not because I was here in Canada. We have learnedthat we can achieve anything as long as we want it and put some effort in doing tomake it happen.Lila my love, thank you too for all those wonderful moments we have sharedthroughout these years. Thanks for all the support you have given me in sadmoments, when I was injured, or feeling lonely. Thanks for being such a greatpartner and best friend. Thanks for patiently helping me with my defense, forhelping me choose the clothes I was supposed to wear on each of my conferencepresentations, and for being there with me no matter what. Let’s keep it up for theyears to come.Homero, you came home so that by the time I moved to Canada you could fillthat empty space. You have provided company, protection, happiness, and evenlots of health benefits for the family. I miss you here but you are loyal and lovedback there at home. Keep it up, you are the best black Lab in the whole world.Woof woof!To all the special people that sadly cannot share this milestone with me any-more, thanks for being there when my family and I needed you. Libia and Lucy,there are not enough words to describe how I miss you. Thank you and yourfamilies for everything. You will always be in my heart.Last but not least, thanks to all my friends from all around the world for be-lieving in my research.xxivTo mom, dad, and Caro.To my loved Lila.To my dog Homero.For the fight against cancer.xxvChapter 1IntroductionNeuroendocrine tumors (NETs) are formed when hormone-producing cells of thebody’s neuroendocrine system begin to grow uncontrollably. These tumors typi-cally originate from such cells in the pancreas, parathyroid, adrenal and pituitaryglands, calcitonin-producing cells of the thyroid, and cells from the gut [3]. NETsoriginating in the gastrointestinal tract and some similar tumors that originate inthe lungs and thymus are usually called “carcinoid tumors” [4]. For the rest ofthis document, no distinction will be made between carcinoids and NETs will beused to refer to all types of neuroendocrine tumors.Yao et al. analyzed prognostic factors for NETs identified in the surveillance,epidemiology, and end results (SEER) database (November 2006) [5]. The SEERdata contains almost 5 million cases of neoplasms in 4,466,501 patients. Theyobserved an increase in the incidence of this type of tumors from 1.09/100,000in 1973 to 5.52 in 2004 per 105 inhabitants. This data suggests that currently NETsare more prevalent than previously reported. However, the therapy outcomes havenot improved. Based on these findings, the US National Cancer Institute and UScongressional committee for the National Institutes of Health Appropriations haveelevated NETs diseases to number two in priority for research funding [1].11.1. Neuroendocrine Tumor Diagnosis1.1 Neuroendocrine Tumor DiagnosisThe diagnosis of NETs is based on:• Clinical presentation: These tumors usually grow very slowly and a highindex of suspicion is required for diagnosis. NETs synthesize, store, andsecrete peptides and neuroamines. The clinical presentation depends onwhether the secreted peptides cause symptoms. Most of these tumors, how-ever, present symptoms that are caused rather by their metastases (e.g. hep-atic metastases) than by the tumor. A late diagnosis, typically 5-7 yearsafter the tumor began to grow, is very common and increases chances ofpresence of metastases. [1]• Hormone assays or identification of biochemical and tissue markers:The identification of biochemical markers found in body fluids is used tocorrectly diagnose NETs. Figure 1.1 shows some of these markers and theiruse. Some markers are common to several types of NETs, while others arespecific to only one type. However, these markers do not provide informa-tion which would allow doctors to predict the growth of the tumor.[1]To be able to provide a better treatment for the patients, an accurate localiza-tion and evaluation of the extent of tumor are required [1, 6]. Medical imagingmodalities used to achieve this purpose include trans-abdominal ultrasound, com-puted tomography (CT), magnetic resonance imaging (MRI), angiography, nu-clear medicine imaging, and intra-operative methods. According to Modlin et al.the location of primary gastroenteropancreatic tumors is not correctly identified21.1. Neuroendocrine Tumor DiagnosisFigure 1.1: Biochemical markers used for diagnosis of NETs. (Copyright Else-vier, taken from Modlin et al. [1] with permission.)in 20% - 50% of the cases [1]. Table 1.1 shows the probability of detection ofpancreatic NETs with different imaging modalities. None of these modalities is100% sensitive and a combination of several of them might be required for anaccurate staging of tumors.Imaging Modality Detection of Pancreatic NETs [%]Ultrasound 13 - 27CT or MRI 22 - 45Selective Angiography 40 - 75Scintigraphy 57 - 77Table 1.1: Probability of detection of pancreatic NETs with different imagingmodalities. [1]31.1. Neuroendocrine Tumor DiagnosisThe introduction of positron emission tomography (PET) and single photonemission computed tomography (SPECT) systems in combination with CT (re-ferred to as PET/CT and SPECT/CT systems) allows physicians to correlate phys-iological and anatomical information. This provides an advantage over singlemodality. Krausz et al. [6] has showed that the use of SPECT/CT improved thediagnosis of NETs compared with planar scintigraphy in 32% of their scintigraphypositive patients, and provided better management in 14% of them. SPECT led tobetter anatomical localization of tumors and improved detection of lesions previ-ously missed on CT. With all this new clinical information available, planning ofthe surgical intervention for these patients also improved.1.1.1 Diagnosis with Nuclear Medicine TechniquesAdams et al. [7] performed a study using fluorine-18 fluorodeoxyglucose to di-agnose NETs using PET. Their results were not very promising as the methodshowed to be useful only for highly aggressive tumors. Amine precursors labeledwith 11C or 18F have also been proposed [8, 9] and have been shown to be moreeffective than CT and scintigraphy.NETs cell membranes possess receptor proteins that have high affinity for reg-ulatory peptides and overexpress somatostatin receptors (SSR) [10]. Somatostatinis a neuropeptide that inhibits secretion of hormones (e.g. growth hormone andinsulin) by binding them to SSR [11]. Based on this property, nuclear medicineimaging techniques, such as SPECT [6] and PET [12] with radiolabeled somato-statin analogs have become widely accepted in staging and characterization of the41.2. NETs TreatmentNETs disease [13]. SSR have been targeted with somatostatin analogues suchas octreotide labeled with 123I [14] or pentetreotide labeled with 111In [12, 15].More recently, somatostatin analogues labeled with 99mTc [16–19] have shownimproved NETs and NET’s metastasis diagnosis compared to other radioisotopes.PET scans using 68Ga [20] are also gaining popularity for NETs diagnosis, but arelimited by PET camera availability. Also, Gallium generators are expensive andnot widely available.1.2 NETs TreatmentDue to wide range of tumor burden and variety of symptoms, NETs treatmentshould be individualized [1]. For some patients, the choice of treatment dependson whether reducing tumor growth is a priority compared to reducing the symp-toms by inhibiting secretion of biological agents or vice versa. Guidelines for themanagement of NETs have been published by Ramage et al. in 2005 and updatedin 2012 [21].1.2.1 Surgery, Chemotherapy, and External Beam RadiationTherapyThe first option in the treatment of NETs is always surgery when tumors areresectable or when palliation can be achieved by debulking [21]. It is usuallythought to be the most effective to control tumor related bleeding and perforation.Also, if the primary lesion is removed, the symptoms generated by the secretion51.2. NETs Treatmentof the bioactive agents can be reduced [22, 23].The decision to perform surgery depends on the exact location and the extentof tumor. In primary pancreatic NETs for example, surgery is the only methodcurrently available that can result in a complete cure [11, 23]. However, becausemost of NETs are diagnosed when they have already metastasized, surgery is oftennot enough to stop the disease. Another treatment option is liver transplantationfor patients with metastatic NETs [24–27]. This practice is still not well acceptedas many patients develop recurring disease.Chemotherapy is another NETs treatment. The introduction of somatostatin(SST) analogues provides control of the symptoms [22] in about 75% of patientsand reduces the concentration of tumor biomarkers [1]. The two most commonagents are octreotide and lanreotide although no significant differences regard-ing quality of life have been observed, lanreotide is preferred due to its modeof administration which requires only one injection every 10 days as opposed tooctreotide which should be administered twice a day [28]. Although these so-matostatin analogues rarely reduce tumor size, symptoms reduction and reductionin tumor markers are common due to the decrease of hormone production thatthese tumors cause [11]. SST have also shown improvement in the progressionfree survival [29].Interferon has also been used in the treatment of NETs [30] reducing symp-toms but not providing a satisfactory tumor response [11]. Combining SST ana-logues with interferon has led to a lower risk of progressive disease compared toSST alone [31], but with some adverse effects.61.2. NETs TreatmentPoorly differentiated NETs are usually first treated with chemotherapy usingdrugs such as capecitabine. Studies regarding the combination of capacetibine andother agents have been reported. A first study combining it with rofecoxib [32]was interrupted because no responses were seen in the patients. Bajetta et al. [33]published a study in which 30% of treated patients showed only a partial responsewhen combining capecetibine with oxaliplatin. Together with temozolomide, thetreatment showed a good response in pancreas carcinomas [34]. In general, con-ventional agents have shown disappointing results with a response rate within 20%- 40% [22, 35]. Recently, the development and testing of new drugs that targetproangiogenic molecules which are overexpressed in NETs shows the potential toadvance the management of the disease [1, 36]. NETs have high vasculature andthus the aim of this novel drugs is to inhibit angiogenesis [11].Regarding external beam radiation therapy, it has been mostly used in welldifferentiated NETs with brain, bone, or spinal metastases which pose a risk ofmyelocompression [11]. Additionally, external beam radiation therapy has beenused to relieve bone pain from metastases, and in some cases in patients withabdominal lesions [11, 21, 37].1.2.2 Peptide Receptor Radionuclide TherapyBesides treatment options mentioned in the last section, peptide receptor radionu-clide therapy (PRRT) has been recognized as an effective tool to treat NETs [38]which cannot be extracted through surgery and/or have metastasized [11]. The factthat these tumors overexpress SST receptors, has led to the development of this71.2. NETs Treatmenttherapy [10]. Somatostatin analogs labeled with therapeutic radionuclides, suchas 111In, 90Y, and 177Lu, have shown very promising results [11, 38]. Commonlyused somatostatin analogues include DOTATOC ([X-DOTA0,Tyr3]-octreotide),DOTATATE ([X-DOTA0,Tyr3]-octreotate), and DOTANOC ([X-DOTA,1-NaI3]-octreotide) [39–45]. The goal of PRRT is to deliver the maximum possible doseto the tumor, while keeping healthy organs safe.The first generation of PRRT used Octretotide labeled with 111In, which emitsγ ′s and Auger electrons. At that time, somatostatin analogues labeled with the betaemitters 90Y and 177Lu were still not available. Several studies [46, 47] investi-gated the therapy with 111In. The results showed that symptoms were improvedbut in general tumor sizes were not reduced. Treatments were carried out withtotal injected activities ranging between 10 GBq and 100 GBq. Side effects weremostly caused by bone marrow toxicity, estimated to occur at dose levels higherthan 3 Gy, as leukemias were induced in three out of 54 patients. One patient wasdiagnosed with renal insufficiency but the authors suggested that it could havebeen caused by a preexisting condition rather than by the treatment itself. How-ever, 111In is not the best isotope to use for PRRT as the range of Auger electronsis very small and tissue penetration is not optimal [11].The second generation of PRRT for NETs involved studies with DOTATOClabeled with 90Y. Several therapies with total injected activity of up to 7.4GBq/m2(normalized by patients surface area) through either two or four cycles were per-formed [48–53]. Some patients received additional amino acid infusions to pre-vent renal toxicity and patients that developed renal insufficiency were those that81.2. NETs Treatmentdid not receive the infusion. Overall, the studies concluded that this type of treat-ment shows promising therapeutic results based on disease stabilization, partialremission, and reduction in tumors [11].The current generation of NETs treatment with targeted radioisotopes involves177Lu labeled DOTATATE. In contrast to 90Y which only emits β particles, 177Lualso emits γ ′s that allow to perform imaging studies for tracking of disease pro-gression. Another advantage of 177Lu with respect to 90Y is its lower range β ’sin tissue due to their lower energy. This allows for better treatment of smaller tu-mors without penetrating healthy tissue. Two publications by Kwekkeboom et al.[54, 55] show that regarding tumor regression, therapies with this agent are verypromising. Patients were treated with cumulative activities of up to 29.6 GBq,using an upper limit of dose delivered to the kidney of 23 Gy. Minor responsewas observed in 19%, partial response in 26%, and complete remission in 2% ina population of 131 patients. If protective agents for the kidneys are given to thepatient, then the side effects of this therapy are minimal. Time to progressionwas also found to be better than with chemotherapy. A study by Teunissen et al.[56] which evaluated the quality of life of patients treated with 177Lu DOTATATE,showed an improvement of symptoms especially in patients with tumor regression[11]. Regarding the radioisotopes, 177Lu seems the most promising one with a sur-vival benefit of several years and an improvement in patients’ life quality whileonly rarely showing any adverse effects [38].The therapies using SST analogues labeled with the three radioisotopes men-tioned above showed satisfactory results. Recent studies have tried to determine91.2. NETs Treatmentwhich of the three peptides (DOTATATE, DOTATOC, or DOTANOC) is better.However, no final conclusion has been reached. The research group from BadBerka [40], analyzed the in-vivo behavior of these three peptides by measuringorgan and tumor kinetics as well as dosimetry. They concluded that the uptakeof DOTATOC was the lowest in healthy tissue, with best tumor to kidney ra-tio. Esser et al. [41], compared DOTATOC with DOTATATE and concluded thatDOTATATE had a longer residence time in both tumors and kidneys. The dose totumors was always higher. Kam et al. [42], confirmed promising results of ther-apies using DOTATATE. These three studies were performed with the peptideslabeled with 177Lu.In parallel, trying to improve diagnosis, Poeppel et al. [43] performed a studyin which they compared PET images of DOTATOC and DOTATATE labeled with68Ga. They concluded that both peptides have comparable diagnostic accuracy,with DOTATOC having “a potential advantage”. Velikyan et al. [44] performedthe same comparison, but they concluded that due to the slight difference inhealthy organ uptake and excretion, DOTATATE is preferable. As no conclusionhas been reached, more studies to test the clinical relevance of different peptidesare needed.1.2.3 Limitations of PRRTA main limitation of PRRT using 177Lu is that currently all patients are treatedwith the same protocol, usually in four cycles with an injected activity of 7400MBq per cycle. This protocol has been created by considering a maximum dose101.2. NETs Treatmentwhich could be delivered to the kidney being equal to 23 Gy [42] or 27 Gy [57].These dose limitations were determined based on toxicity levels derived from ex-ternal beam radiation therapy studies.However, radiotracer uptake in tumors and healthy tissue has been observedto differ greatly between patients [20] indicating that a big group of patients couldtolerate a higher dose which possibly could improve therapy outcome. Bodei et al.[39] showed that patients with risk factors such as diabetes and hypertension havebiological effective dose thresholds to the kidney equal to 28 Gy, but for patientswithout those risk factors, kidney toxicity had a higher threshold of up to 40 Gy.To summarize, the problem with the “one dose fits all approach” in currentNETs PRRT is that some patients are being left under-treated, while others may beover-treated. Individual responses and toxicities fluctuate largely from patient topatient. It is generally believed that treatment plans using injections that would bebased on an individualized radiation dose assessment could significantly improvePRRT outcomes and should become routine as it is done in the external beamradiotherapies [58]. This however, requires an accurate dose estimate deliveredto healthy organs and tumors, which is believed to be a very difficult and timeconsuming procedure.Personalized dosimetry requires the accurate determination of the activity dis-tribution within organs of interest, its behavior over time, and physical informationof the radioisotope being used. The temporal behavior can be measured by usinga series of nuclear medicine scans performed at different time intervals after theinjection. Quantitative corrections for physical effects that degrade image qual-111.3. Outline of the Thesisity should be applied when performing image reconstruction of the data acquiredwith SPECT/CT cameras [59, 60].This project was focused on developing a protocol for accurate quantifi-cation of 177Lu in order to perform personalized dosimetry calculations inthe treatment of NETs. It was conducted at the Medical Imaging ResearchGroup (MIRG) of the University of British Columbia in Vancouver, Canada.1.3 Outline of the ThesisOur investigation of methods which would allow us to achieve quantitative resultsfor SPECT/CT of 177Lu for future PRRT personalized dosimetry, were dividedinto three parts:1. Lutetium-177 is a β emitter. Electrons produce electromagnetic radiationwhen interacting with atoms of the medium through a process known asBremsstrahlung (BRS). The first part of the study was to investigate the be-havior of BRS radiation produced by β particles emitted by 177Lu in tissue,in order to understand its contribution to the energy spectrum of photonsdetected by the SPECT camera.2. An essential requirement for accurate personalized dosimetry is the cor-rect determination of radioactivity in organs and other regions of interestwithin a patient. Therefore, the second step of this project consisted in de-termining the accuracy of activity quantification that could be achieved in121.3. Outline of the Thesis177Lu SPECT/CT imaging studies. A series of phantom experiments wasperformed and two scatter correction methods were compared.3. The next step involved corrections for photons that are not recorded by theSPECT camera when count rate is very high (i.e.corresponding to values ofactivities equal or higher than 7 GBq). This effect is known as camera dead-time (DT), and experiments to measure DT under conditions correspond-ing to patients scans in a clinical environment were performed. Withoutthis crucial correction, the dose delivered to the patient would be underesti-mated.After these steps were completed, the findings were applied to patient data ob-tained from our collaborators in Quebec, and dosimetry calculations were per-formed for these patient’s kidneys. Finally, a computer software (graphical userinterface (GUI)), SPEQToR, was developed. Our objective was to make our quan-titative image reconstruction software easy to distribute and easy to operate ina clinical environment. The software is capable of reconstructing quantitativeSPECT/CT data and perform dosimetry with deadtime corrections.This thesis contains eight chapters. Chapter 2 provides a brief background onthe basic theory required to generate tomographic images, types of corrections tomake those images quantitative, presents the way in which internal dosimetry isperformed, and gives some details on the radioisotopes used in PRRT, focusingon 177Lu, its production, and the interaction of particles with matter (tissue).131.3. Outline of the ThesisChapter 3 deals with Monte-Carlo (MC) simulations of 177Lu sources in waterin order to understand BRS contributions to the detected energy spectrum. Chap-ter 4 describes the phantom experiments performed with low activities and pro-vides quantification accuracies for our protocol. Chapter 5 discusses the deadtimeeffect occurring at count rates typical for therapy and our proposed correction forit. Chapter 6 describes a complete application of our dosimetry protocol to data offour patients treated with 177Lu-DOTA-octreotate. These data was obtained fromour collaborators from the Department of Medical Imaging at Universite Laval,Quebec City. Chapter 7 shows the developed graphical user interface for quan-titative reconstruction and its tools necessary to perform dosimetry. Chapter 8summarizes all the important conclusion obtained from this work, states its con-tributions to the field, and discusses ideas for future work necessary to improvePRRT for personalized dose assessment in the clinical environment.The chapters were written following a scientific publication scheme includinga self contained introduction, methods, results, discussion and conclusion.14Chapter 2Background of Nuclear Medicineand Dosimetry2.1 What is Nuclear Medicine?Nuclear medicine (NM) is a medical specialty that uses radioisotopes to diagnoseand/or treat diseases. In order to do this, compounds labeled with radionuclides,known as radiotracers or radiopharmaceuticals, are introduced into the patient viaan injection or oral administration. The radioactive decay of the radionuclide usedin diagnostic imaging studies creates gamma rays either in a direct or indirect way.These gammas must have enough energy to exit the patient’s body. By placinggamma detectors around the patient, it is possible to measure emitted photonsand generate images of the distribution of the radiolabeled compound within thepatient. [61]NM has two main categories for imaging:1. Single photon imaging: Some isotopes emit many photons. If images aregenerated taking into account individual photons the modality is known assingle photon imaging. If two-dimensional (2D) images are generated, then152.1. What is Nuclear Medicine?the technique is known as scintigraphy or planar imaging.This is done byplacing the detector in one position about the patient. If three-dimensional(3D) images are generated, the technique receives the name of single pho-ton emission computed tomography (SPECT) in which several planar viewsobtained at different positions around the patient are combined to form a 3Dimage. SPECT images are preferred because they provide depth informa-tion, better contrast than planar scintigraphy, and images do not containoverlapping organs, thus are better suited for the accurate assessment ofpatient’s health. However, data acquisition is usually longer compared toplanar imaging. [61, 62]2. Positron imaging: This technique deals with radiotracers that emit positronswhen they decay. The positrons annihilate with electrons in the vicinity ofthe place where they were produced creating two gammas with an energyof 511 keV each that travel in almost completely opposite directions. Thesephotons are detected at different angles around the patient, and tomographic(3D) images are generated. The technique receives the name of positronemission tomography (PET). [61, 62]Other imaging modalities such as MRI, imaging with X-rays (both planar or 3D)are very good at providing anatomical information but have limitations if biologi-cal or physiological information is required. While NM techniques can detect ra-diolabeled substances of concentrations of nanomolar or picomolar, methods thatinvolve magnetic resonance can only detect substances in the millimolar concen-162.2. Nuclear Decaytrations [61]. This very high sensitivity makes NM the technique of choice whenthe understanding of disease down to its molecular level is desired. Examples ofthis are tissue perfusion, glucose metabolism, and identification of receptors intumor cells. However, the radiotracer should have three main characteristics inorder to provide the information in the way it was planned.• The chemical properties of the injected compound must not be changed byradioactivity.• The radioactive properties of the radioisotope must not be changed by chem-ical reactions.• The radiotracer should not disturb the biological processes occurring withinthe patient.The design of the radiotracer depends on the physiological aspect under inves-tigation. Some radiotracers are analogous to substances that are already presentin the body, like glucose, while others are designed to bind to specific sites likedopamine receptors present in the brain or SST receptors in NETs cell membranes.[62]2.2 Nuclear Decay2.2.1 Decay EquationsNuclear or radioactive decay is a random event that occurs in any substance thatcontains unstable isotopes that are constantly transforming in time. If the number172.2. Nuclear Decayof unstable nuclei is N(t) at time t, and the probability of decay in an interval oftime dt is given by λdt, then the rate of change of the number of nuclei, known asthe activity of the sample A(t) is equal to:A(t) =−dNdt= λN(t) (2.1)By solving this differential equation, the number of remaining nuclei (thathave not decayed) or the remaining activity, at a time t can be calculated:N(t) = N0e−λ tA(t) = A0e−λ t(2.2)where the initial number of nuclei and initial value of activity are representedby N0 and A0, respectively. The constant λ receives the name of the decay con-stant, has units of 1/time, and is characteristic for each radioactive decay of aradioisotope. A very common way of characterizing a sample, is by calculatingthe time it takes to decay to 50% of its initial value (number of atoms or radioac-tivity), and it is related to λ byT1/2 =ln(2)λ(2.3)where T1/2 is known as the half life of the isotope.The SI unit for activity is the Becquerel (Bq) and is defined as 1 decay persecond. However, another very common unit for radioactivity is the Curie (Ci)and is related to the Bq as 1Ci= 3.7×1010Bq . [61–64]182.2. Nuclear Decay2.2.2 Decay ModesNuclei can decay by several modes that include emissions of alpha (α), beta (β ),gamma (γ), and electron capture (EC). Before getting into the details of each de-cay mode, it is important to understand the notation used to label nuclear species.This is given by:AZXNwhere• X represents the symbol of the element as shown in the periodic table.• Z is the number of protons and is known as the atomic number.• N is the number of neutrons.• A is the mass number, which is given by A= Z+N.Several characteristics in nuclide composition define the nuclear families that aresummarized in Table 2.1. [62, 64]2.2.3 Alpha DecayErnest Rutherford found that some heavy nuclei were emitting particles that didnot penetrate deep in materials and were positively charged. These particles are192.2. Nuclear DecayFamilyName Characteristics ExamplesIsotopes Nuclides with the sameproton number ZA1Z XNA2Z XNA3Z XN17771 Lu10617671 Lu10517571 Lu104Isobars Nuclides with the samemass number AAZ1XN1AZ2YN2AZ3ZN317771 Lu10617772 H f10517780 Hg97Isotones Nuclides with the sameneutron number NA1Z1 XNA2Z2 YNA3Z3 ZN17771 Lu10617670 Yb10617165 Tb106Table 2.1: List of nuclear families with some examples.very ionizing but can be stopped by a sheet of paper. They were identified asHelium atoms and the α decay follows the equation:AZXN →A−4Z−2 YN−2+42 He2+Qα (2.4)Alpha decay is dominant in heavy nuclei (A≥ 210) and the energies of emittedalpha particles range from 4-9 MeV. [63]2.2.4 Beta DecayThis type of decay is the most common and has been detected in isotopes of manyelements.1. β− decay: In this case, one of the a neutrons (n) in the nucleus is trans-formed into a proton (p+) , an electron (e−) (which is the β− particle), and202.2. Nuclear Decayan antineutrino (ν¯). Because the available decay energy is split betweenthe two particles, the β− energy spectrum of this decay is continuous. Thisdecay generates a different daughter element than the parent and is knownas a transmutation decay. This decay is only allowed if the mass of the par-ent atom is greater than the mass of the daughter. The β− is given by thefollowing equation: [62–64]n→ p++β−+ ν¯e+ energy (2.5)or in nuclear notationAZXβ−−−→ AZ+1Y (2.6)2. β+ decay: In this case, one proton in the nucleus is transformed into aneutron, a positron (e+) (the β+ particle), and a neutrino. It follows thefollowing equation:p+→ n+β++νe+ energy (2.7)or in nuclear notationAZXβ+−−→ AZ−1Y (2.8)This decay is again a transmutation decay because the daughter nucleus isa different element than the parent. This decay is only allowed if the massof the parent atom is greater than the mass of the daughter plus two electron212.2. Nuclear Decaymasses. [62–64]3. Electron Capture: Is a process that competes with β+ decay. In it, anelectron from an inner atomic shell is captured by the nucleus, a proton istransformed into a neutron and a neutrino is ejected from the nucleus. Itfollows the following equation:p+ e−→ n+ν+ energy (2.9)or in nuclear notationAZXEC−−→ AZ−1Y (2.10)The parent nucleus follows a similar transmutation as in the β+ decay, but inthis case the decay is only allowed if the mass of the parent is greater than the massof the daughter plus the electron ionization energy. After the electron is captured,another electron from an outer shell fills the gap emitting a photon (characteristicx-ray), or the energy is used to eject another orbital electron. The latter case isknown as an Auger electron. [62–64]2.2.5 Gamma DecayIn gamma decay, a nucleus in an excited state, denoted with an asterix (∗), releasessome of its energy by emitting a photon. This decay is not a transmutation and itfollows the equation:222.2. Nuclear DecayAZX∗→AZ X+ γ (2.11)The de-excitation can occur from an excited state to the ground state or fromone excited state to another. [63]A process competing with γ emission is emission of conversion electrons. Inthis process, the excited nucleus can also interact with the electrons of the atomand the energy is released by ejecting an electron rather than by emitting a photon.This process is known as internal conversion.2.2.6 177Lu Decay and ProductionLutetium-177 is a β− decaying isotope with a 6.6 days half-life. Beside β ′s, italso emits gamma rays. The reason for this is that the β decay of 177Lu leadsto several excited states of 177Hf which consequently decay by a gamma decay(and electron conversion). The partial decay scheme of this isotope is shown inFigure 2.1. The energies of the most abundant gammas are 113 keV and 208 keVwith intensities of 6.2% and 10.3%, respectively. Other gammas at 249 keV and321 keV are also present but with lower intensities. The maximum and meanβ energies of 177Lu decay are 498.3 keV and 134.2 keV, respectively [65], withcorresponding maximum and mean ranges of these β particles in tissue of 2.0 mmand 0.5 mm [66], respectively.177Lu is produced in a nuclear reactor by one of two methods:1. Direct production through the 176Lu(n,γ)177Lu reaction by neutron capture232.2. Nuclear Decay177Lu-:100%0113249.7321.30112.9208.4249.7321.37/2-9/2-11/2-9/2+177HfFigure 2.1: Decay scheme of 177Lu into 177H f .on an enriched 176Lu target (>68.7% [67]) [68, 69].2. Indirect production using an enriched 176Yb target that follows the reaction176Yb(n,γ)177Yb→177 Lu. The intermediate product, 177Yb in this case,has a half-life of 1.91h and decays to 177Lu [69].Theoretically, the specific activity obtained from the direct production is lowerthan the one from the indirect method due to the competition with the 176Lu(n,γ)177mLu[68]. However, the separation of 177Lu from the Yb2O3 target is difficult and thepresence of Yb reduces the specific activity [69].Table 2.2 summarizes the decay characteristics of 177Lu and compares then toother isotopes used in radionuclide therapies.242.3. Interaction of Radiation with MatterIsotope Half-life γ Energies [keV](Abundance [%])Max βEnergy[keV]Mean βEnergy[keV]Max βRange inTissue[mm][66]Mean βRange inTissue[mm][66]177Lu [65] 6.6 d113 (6.2),208 (10.3) 498.3 134.2 2.0 0.5188Re [70] 17.0 h155 (15.6),478 (1.1),633 (1.4)2120.4 763 11.0 3.890Y [71] 64.1 h - 2280.1 933.6 11.3 4.1Table 2.2: Decay properties for 90Y, 188Re, and 177Lu (only gamma emissionswith abundance higher than 1% are shown).2.3 Interaction of Radiation with MatterAlthough there may be more ways in which radiation can interact with matter, thissection will only focus on those relevant to the decay of 177Lu. As this isotopedecays via the β− and emits gammas, only interactions of these two types ofradiation will be discussed.2.3.1 β Particles InteractionsThe β particles emitted by 177Lu, slow down as they travel through matter bycolliding with atoms and molecules. Several outcomes can occur: [61, 63, 64]1. Ionization: If the β particle interacts with an orbital electron of an atom, theβ particle loses some of its energy by transferring it to the orbital electron.If the energy transferred to the orbital electron is higher than its bindingenergy, the electron will be ejected from the atom. The vacancy will befilled by an electron from an outer shell. When this happens, either a pho-ton (characteristic x-ray) will be emitted, or the energy difference between252.3. Interaction of Radiation with Matterenergy levels would be used to eject another electron from an outer shell (anAuger electron). Some of the ejected electrons, known as δ -rays, can haveenough energy to cause secondary ionizations. [61, 63, 64]2. Excitation: If a β particle interacts with an orbital electron, but the energyis transferred in a way in which the electron is raised to an excited state theprocess is known as excitation. In this case, the β particle usually loses lessenergy than in the ionization case. [61, 63, 64]3. Bremsstrahlung: If the β particle penetrates the orbital cloud of the atomand interacts with the nucleus, it is deflected by the electric field generatedby the nucleus. The deceleration causes the β particle to lose some energywhich is carried away by a BRS photon. The energy of the photon can bevery small, when the β particle is only slightly deflected from its originaltrajectory, or be equal to the total energy of the β particle, in which case theparticle is basically completely stopped by the electric field. [61, 63, 64]The first two cases are known as collisional energy losses, while the third case isknown as radiation energy losses. The total energy loss (dEdx )Tot of an electron canbe divided into the collisional and radiation components:(dEdx)Tot = (dEdx)coll+(dEdx)Rad (2.12)The energy losses due to BRS increase with the increasing energy of the βparticle, and with the atomic number Z of the material through which the particleis traveling. [61, 63, 64]262.3. Interaction of Radiation with MatterIncident gammaPhotoelectronFigure 2.2: Diagram of photoelectric effect. An incident gamma is absorbed bythe atom and is used to eject an electron.2.3.2 Gamma InteractionsElectromagnetic radiation (the gammas emitted from the decay of 177Lu and thephotons generated by the interactions of β particles with the medium) interactwith matter mainly by three processes:1. Photoelectric Effect: In this process, the gamma is absorbed by an atomand its energy is used to eject an electron. The energy of the emitted electron(Ee) given byEe = Eγ −Be (2.13)where Eγ is the initial energy of the gamma and Be is the binding energyof the electron. This effect mostly occurs on inner shell electrons. Thephotoelectric effect dominates the interactions of photons at low energiesand is proportional to Z4/E3γ . A diagram of the photoelectric effect is shownin Figure 2.2. [61, 63, 64]2. Compton Scattering: This effect occurs when a photon interacts with afree electron or an electron weakly bound to the atom. Due to the much272.3. Interaction of Radiation with Matterhigher energy of the gamma compared to the binding energy of the electronin the atom, the interaction can be assumed to occur between the photonand a free electron. Contrary to photoelectric effect, where the photon com-pletely disappears, in this case a secondary photon is emitted at a certainangle θ . If the energy of the initial photon was E0, then the secondary(scattered) photon energy isE ′ =E01+ E0mec2 (1− cos(θ))(2.14)where mec2 is the rest energy of the electron. The electron gains a kineticenergy Ee equal to the difference between the energies of the initial and thescattered photon.Ee = E0−E ′ (2.15)For θ = 0, where the photon is not deflected, E0 = E ′, while for the case inwhich a photon is back-scatter θ = 180◦, E ′ = E0/(1+ 2E0mec2 ), the deflectedphoton carries the least amount of energy. As Compton effect is assumedto occur on free electrons, the probability of this interaction depends on thenumber of electrons per gram and not on the atomic number Z. The proba-bility of a photon to scatter at a certain angle is given by the Klein-Nishinaformula which provides the differential cross section for this process as afunction of scattering angle and the initial energy of the photon. A diagramof Compton effect is shown in Figure 2.3. [61, 63, 64]282.3. Interaction of Radiation with MatterIncident gammaCompton electronScattered gammaFigure 2.3: Diagram of the Compton effect. An incident gamma interacts with aweakly bound electron and scatters. The scattered photon has energy lower thanthe incident photon.3. Rayleigh Scattering: In this effect, the photon, instead of scattering onan electron, scatters on the atom as a whole. As the atom has mass muchbigger than an electron, there will be no energy transfer to the atom, and thephoton is scattered without losing energy. This effect is important at verylow energies (<50 keV), usually not of importance for NM applications.Although pair production is a fourth process by which photons interact , it is notrelevant for the energy ranges of 177Lu and thus is not discussed here.Each of the gamma interacting processes has its own probability of occurrencethat is dependent on the photon energy and the composition and thickness of themedium through which the photon travels. When the photon interacts, it is saidthat the photon is attenuated. The number of attenuated photons dN, if N photonsare incident on a medium of thickness dx follows the equationdN =−µNdx (2.16)292.3. Interaction of Radiation with Matterwith µ known as the linear attenuation coefficient and is usually expressed incm−1. The attenuation coefficient is proportional to the medium density ρ , how-ever, to remove density effects µ may be expressed as the mass attenuation coef-ficient µm in cm2/g. whereµm =µρ(2.17)The value of the mass attenuation coefficient depends on the energy of the inci-dent photon and the atomic number of the material. The total µm is the sum ofindividual mass attenuation coefficients corresponding to each of the possible pro-cesses in which photons can interact with matter. Figure 2.4 shows µm behaviorfor water, lead, and NaI for photoelectric effect, Compton scatter, Rayleigh scat-ter, and pair production (although non-existent for these energies) as a function ofincident photon energy. The energy range has been set from 0 to 500 keV as it isthe range relevant to 177Lu SPECT procedures.By solving equation 2.16 , the number of photons remaining after passingthrough a certain depth of material (in cm for linear attenuation and g/cm2 for thecase of mass attenuation coefficient), can be expressed asN = N0e−µx (2.18)where N0 is the initial number of photons that are incident on the material.[61, 63, 64]302.3. Interaction of Radiation with Matter1 10 10010-810-710-610-510-410-310-210-1100WaterMass Attenuation Coefficient [cm2/g]Photon Energy [keV](a)1 10 10010-210-1100101102103LeadMass Attenuation Coefficient [cm2/g]Photon Energy [keV] Rayleigh Compton Photoelectric Pair Production Total(b)1 10 10010-210-1100101102103104Mass Attenuation Coefficient [cm2/g]Photon Energy [keV]NaI(c)Figure 2.4: Mass attenuation coefficients for water, lead, and NaI asa function of photon energy. (The data to generate these plots wasobtained from the National Institute of Standards and Technology web-page http://physics.nist.gov/PhysRefData/Xcom/html/xcom1-t.html on Novem-ber 2015). 312.4. Nuclear Medicine ImagingFigure 2.5: Picture of a Siemens SymbiaT SPECT/CT camera system. This cam-era has two detectors attached to a gantry. They can rotate around the patientin order to collect projections at different angles. An x-ray tube is also presentallowing for the collection of anatomical data using CT.2.4 Nuclear Medicine Imaging2.4.1 The Gamma CameraIn order to be able to detect gammas coming from the interior of the patient af-ter the injection of a radiotracer, a gamma camera is used. Figure 2.5 shows aSiemens SymbiaT SPECT/CT camera. The camera is composed of a collimator,an NaI crystal detector, a back compartment where photomultiplier tubes (PMT)’sand circuits are included, a gantry, a patient bed, and a computer. [61, 64, 72]322.4. Nuclear Medicine ImagingCollimator:The collimator is a lead (Pb) plate with holes that allows only gamma rays travel-ing from certain specific directions to reach the detector. Collimators are charac-terized by their sensitivity and resolution. For parallel hole collimators, the sensi-tivity which is the number of events (counts) recorded by the camera per minuteper unit of activity of the source (cpm/kBq), is independent of the distance fromthe source to the collimator. The collimator resolution, on the other hand, doesdepend on the distance from the source to the collimator and the geometry of thecollimator. Depending on their purpose, collimators are designed with differentseptal thicknesses, hole diameters, and hole lengths. Typical collimators includethe Low Energy High Resolution (LEHR) which is used in the imaging of lowenergy gammas (~140 keV), the medium energy (ME) used for imaging higherenergy photons, and the High Energy (HE) used for even higher energy photons.[61, 64, 72]Crystal:The crystal is the component of the camera that is immediately behind the colli-mator. In typical SPECT cameras, it is made of a sodium-iodide (NaI) scintillator.When the gammas (or β ′s) reach the detector, they interact with the molecules ofNaI. The molecules of the crystal are excited and/or ionized. In NaI (and othermaterials known as scintillators) the energy of these excitations is released aslight. The amount of light emitted after the interaction of the gammas (or β ′s) is332.4. Nuclear Medicine Imagingproportional to the energy deposited by the incident photon within the detector.[61, 64, 72]Back CompartmentThe back compartment of the camera includes PMT and the associated electronics.• The PMT’s generate an electrical pulse from the light generated by the scin-tillator crystal.• The electronic components amplify and shape the electric signal and gener-ate 2D histograms of the location of photon interactions within the crystalwhich allow us to generate images.The camera is usually well shielded in order to avoid recording background radi-ation.2.4.2 SPECT/CT AcquisitionIn case of SPECT, where a 3D image is desired, it is important to collect severalprojections of the object at different angles around it.Modern cameras allow to specify the following acquisition parameters:• Matrix size: Specifies the dimensions of the 2D histogram in the projection.The typical matrix size of a SPECT image is 64×64 or 128×128, meaningthat the 2D histogram will have 64 or 128 bins in each direction.342.4. Nuclear Medicine Imaging• Number of projections: This parameter specifies how many projections willbe collected around the desired object.• Frame duration: This is the time for which the data were collected in oneprojection.• Window setup: The energy range of acceptance of gammas detected by thecamera.• CT acquisition: SPECT/CT cameras (like the SymbiaT shown in Figure2.5) also have a CT (X-rays) acquisition system to collect anatomical infor-mation. Usually the CT is collected before or after the SPECT acquisitionis finished and the patient is still lying on the bed. The importance of CTis that it provides maps of µ values (equation 2.18) for the object beingscanned. In the case of patients, it provides the maps of attenuation coeffi-cients for different tissue types. As the µ values depend on energy, the CTnumbers must be translated to µ to correspond to the energy of the primaryphotons.2.4.3 Image ReconstructionAfter a series of projections of the object are obtained, image reconstruction mustbe performed in order to create a 3D volume of the distribution of activity. Thereare several algorithms [73] which reconstruct the image either analytically or in aniterative way. For the purpose of this project, only the ordered subsets expectationmaximization (OSEM) algorithm [74] will be explained which is an accelerated352.4. Nuclear Medicine ImagingFigure 2.6: Flow diagram of the OSEM algorithm. An initial estimate of the imageis provided and updated on each iteration until the desired image is obtained.version of the maximum likelihood expectation maximization (MLEM) algorithm[75] (see the flow chart presented in Figure 2.6).MLEM was derived by maximizing the likelihood that the reconstructed im-age generates the measured projection data, taking into account that the countsdetected in the projections follow Poisson statistics. By taking groups of mea-sured projections, known as subsets, instead of the complete set of projections,the OSEM is faster than MLEM and follows the equation:X l+1j =X′lj∑i∈ΩiCi j∑i∈ΩiCi jYi∑kCikX lk(2.19)The variables in this equation represent the following:• Yi is the ith measured projection.• X lk is the voxel k of the image estimate X for the lth iteration. The firstestimate has the same dimensions of the reconstructed image and it usuallycontains a value of 1 in each of its voxels.362.4. Nuclear Medicine Imaging• Cik is the i,k element of the system matrix. In general, this matrix containsinformation about the geometry of the system and the process of data ac-quisition. In imaging terms, the value of Cik represents the probability thata gamma that was emitted from voxel k contributes to the counts in a pro-jection pixel i. The greatest advantage of iterative reconstruction methodscompared to analytical is that the system matrix can include corrections forthe physical processes involved.• X l+1j is the new (l+1) estimate of the 3D image.• The iterations are performed on a number of subsetsΩi that contain sub-setsof projections. As an example, assuming that 12 projections were measured,3 subsets containing 4 projections each can be generated containing thefollowing projection numbers:– Ω1 = {1,4,7,10}– Ω2 = {2,5,8,11}– Ω3 = {3,6,9,12}and the iterations are done using the projections of each of these subsetsinstead of using all the projections as MLEM does. This makes OSEM afaster reconstruction algorithm.The steps of the OSEM algorithm are (see Figure 2.6):1. From the image estimate X lk (a matrix filled with ones for the first estimate),using the system matrix projection estimates are created.372.5. Dosimetry2. A projection calculated by taking the ratio of measured projectionYi and theestimated projection obtained from step 1 is determined.3. The modified projection obtained from step 2 is back-projected (i.e. 2Dprojection to 3D image) using the system matrix by ∑i∈ΩiCi jYi∑kCikX lk.4. This image is normalized by ∑i∈ΩiCi j and then multiplied by X′lj in order toobtain the new estimate X l+1j .After several iterations, the result is a reconstructed 3D image.2.5 DosimetryThe Committee on Medical Internal Radiation Dose (MIRD) published a protocolto perform 3D dosimetry in internal radionuclide therapy as the MIRD pamphletnumber 23 [60]. Absorbed dose is the mean energy deposited in a target tissue ororgan per unit mass of the target (tissue or organ). This dose D(rT ,TD) absorbedby the target rT over a dose-integration period TD from the radioactive source rsis given byD(rT ,TD) =∑rsA˜(rs,TD)S(rT ← rs) (2.20)A˜(rs,TD) is the cumulative activity or integrated activity over the time intervalTD and S(rT ← rs) is the factor representing the dose deposited in the target or-gan per nuclear decay in the source organ. A˜ depends on the biologic distributionof activity within the patient, and the S-factor contains details about the physi-382.5. Dosimetrycal properties of the radionuclide. Combining the biologic information with thephysics of the isotope decay is what generates the dose estimate.For the remainder of this chapter, the focus will be on determining an accurateestimate of the activity distribution over time in order to determine the cumulativeactivity (equation 2.20).As discussed in section 2.2.1, equation 2.2 shows that the activity decreasesexponentially with time and that the half life is related to the decay constant asshown in equation 2.3. However, when a radiopharmaceutical is injected intothe patient, the activity not only decays by pure physical properties, but also themetabolism of the patient will eliminate (decrease) the amount of activity in anyorgan. In general, the effective half-life of a radioisotope is given by1T1/2e f f=1T1/2Phys+1T1/2Bio(2.21)with the physical component T1/2Phys and a biological one T1/2Bio . Determi-nation of T1/2e f f presents a challenge in the calculation of A˜ as the biologicalcomponent of the half-life can vary from patient to patient or organ to organ.Therefore, because T1/2Bio is an unknown, patients have to be scanned at differenttime points and the activities for different regions of interest fitted to models suchas a mono-exponential:A(t) = A0e− ln(2)T1/2e f ft(2.22)More complex cases in which an uptake of activity occurs during the first392.5. Dosimetry~Activity [MBq]Time [days]A(t)=A0e-(ln(2)/T1/2)tA=Area Under The Curve(total number of emitted particles)Figure 2.7: Activity washout within a patient. The effective half life has to bedetermined in order to be able to find the total number of emitted particles inorder to perform dosimetry calculations.minutes after injection a biexponential fit might be more accurate. No matterwhat type of fit is used, the goal is to find the area under the curve A˜ in order toperform the dose estimation (Figure 2.7).2.5.1 Quantification of ActivitySo far, the images that are reconstructed following the procedure discussed insection 2.4.3 take into account only photons that reach the detector. There arehowever, several effects, such as attenuation of photons, scattered photons, andcamera deadtime that cause image degradation and prevent an accurate estimateof the activity distribution within the object. However, corrections for these ef-fects are feasible, and once done, quantitative images of activity can be obtained.This means that the distribution of activity within the object can be correctly de-termined.402.5. DosimetryNon-attenuatedPhotonAttenuatedPhotonFigure 2.8: Diagram of attenuation of photons within the object being scanned.Attenuated photons do not reach the detector and this translates into an underesti-mation of activity if this effect of attenuation is not corrected for.Attenuation CorrectionAttenuation of photons was briefly described in section 2.3.2. The effect thatthis has on quantification, as shown in Figure 2.8, is that some photons that areemitted from the object being scanned, are absorbed in the medium and neverreach the detector. Fewer counts in the projection will result in an underestimationof the activity of the source. Equation 2.18 gives the number of photons that havenot been attenuated after passing through a thickness x of a material with theattenuation coefficient µ . The goal of the attenuation correction is to recover theinitial number of photons emitted from the source, by knowing the number thatreached the detector.412.5. DosimetryThis can be done by solving for N0 such thatN0 = Neµx (2.23)The attenuation coefficient value must be determined in order to find a solutionto the equation.The values of µ can be easily determined using CT. These images are com-monly known as µ-maps, and contain the µ value distributionAs in a patient, the distribution of attenuation values is not uniform, the at-tenuation correction procedure is performed in small steps along the directionperpendicular to the detector (projection), so equation 2.23 can be generalized asN0 = Ne∑i µil (2.24)Attenuation correction can be incorporated into the system matrix C of theOSEM algorithm, and the reconstructed image will now display counts emittedfrom the source.Scatter CorrectionAnother effect degrading the accuracy of projections is scatter of photons. Figure2.9 shows an example of a primary (non-scattered) and scattered photon. Al-though both of them are detected, scattered photons provide wrong informationabout the activity distribution. The scattered photon appears to originate from adifferent position than where the source is really located. Therefore in order to422.5. DosimetryPrimaryPhotonScatteredPhotonFigure 2.9: Diagram of the scatter of photons that degrade image quality. Pho-tons can scatter within the object and be detected giving the impression that theyare originate at a different location from where the source is located, and it alsocontributes to a higher number of counts which is translated into higher activityvalues.obtain accurate distribution of activity, scattered photons should be removed fromthe projections, and only primary photons should be used in the reconstructionprocess. Figure 2.10 shows an energy spectrum for 177Lu in which the primaryand scattered photons components have been shown separately.The first method of removing scattered photons from the projection data isby setting energy windows to accept only photons whose energies fall within thelimits of the window. As mentioned in the description of Compton scattering insection 2.3.2, primary photons will lose energy when they scatter. In order toobtain accurate quantification of activity the energy window should be set aroundthe isotope’s photopeak. This however does not remove self-scattered photons432.5. DosimetryEnergy [keV]0 100 200 300 400 500Counts0246810121416310×Figure 2.10: A typical detected spectrum of 177Lu is shown in a dashed line. Thespectrum is composed by the primary photons (red) and scattered photons (blue).that can still have energies within the energy window. The non-removal of thesescattered photons leads to the overestimation of the activity within the object.Although many scatter correction methods have been proposed [76–81], onlythe triple energy window (TEW) [79] and the analytical photon distribution inter-polated (APDI) [80, 81] developed by the MIRG will be discussed here.The TEW method consists of specifying three energy windows, the photopeakand two windows beside it (one on each side of the photopeak). The total countscollected in the photopeak window Cpw are the sum of the primary counts Cprimand scattered counts CscatCprim =Cpw−Cscat (2.25)The scattered counts are estimated using the counts in the lower window Clwand the upper windowCuw. All these windows have a width of wlw, wpw, and wuwfor the lower, photopeak, and upper window, respectively. The scattered counts442.5. Dosimetryare estimated as:Cscat =(Clwwlw+Cuwwuw)wpw2(2.26)APDI on the other hand, uses an estimate of the source distribution and theattenuation map to obtain the distribution of scattered photons up to second orderscatter (i.e. a photon that scattered twice). It analytically calculates the probabilitythat photons emitted from a certain point in the object will scatter within the objectand then be detected at a certain pixel on the projection. The scattered distributionis then estimated by scaling this probability by the number of photons emitted ateach location using the estimate of the activity distribution.When the scatter projections are estimated either with APDI or with TEW,they are subtracted from the measured photopeak window projections creating anew set of projections composed of only primary photons.Camera DeadtimeThe crystal, the PMT’s, and the circuits that process the electric pulses take acertain amount of time to process individual events. If a second event occurringduring the time in which the first one is being processed makes the system inca-pable of processing either, then the system is said to be paralyzable. If a secondevent during the time in which the first is being processed, the system is inca-pable of processing either, then the system is said to be paralyzable. If a secondevent occurs while the first one is being processed but the system only ignores thesecond event, the system is said to be non-paralyzable. In both cases, events are452.5. Dosimetrybeing ignored during a certain amount of time known as deadtime τ . If gammasare reaching the detector at a true count rate Rt and the observed count rate is Ro,then the observed count rate based on the paralyzable model follows equationRo = Rte−Rtτ (2.27)and the observed count rate based on the non-paralyzable model is given byRo =Rt1+Rtτ(2.28)In an ideal system, the true count rate should be the same as the one beingmeasured. Figure 2.11 shows plots of Ro vs Rt . It can be observed that the higherthe number of gammas reaching the detector per unit time, then the deviationfrom the truth (Ro = Rt) increases. Ignoring events due to deadtime leads to anunderestimation of the activity in the scanned object. It is important then to knowthe relation between the observed and true count rates in order to account for thephotons that are missing to obtain accurate activity quantification. [61]Sensitivity/Calibration FactorOnce corrections for attenuation and scatter have been applied, a voxel in thereconstructed image provides information about the number of primary photonsemitted from the object. In order to convert these counts into absolute values ofactivity (i.e. MBq, kBq, Ci, etc) it is important to determine a camera calibrationfactor ε which relates the two quantities. The procedure to obtain this factor462.5. Dosimetry0 1 2 3 401234x105Observed Count Rate [cps]True Count Rate [cps] Paralyzable Model Non-Paralyzable Model Identityx105Figure 2.11: Behavior of camera count rate detection as a function of the truecount rate reaching the detector. The ideal situation in which the observed countrate is the same as the true count rate is shown in blue. Deviations from this linedue to the paralyzable (black) and non-paralyzable (red) models are also shown.The plots have been made assuming a value of τ = 5µs.involves a planar or SPECT scan of a source with known activity. By measuringthe counts (N) in an interval of time (tcal), and knowing the true activity of thescanned source the calibration factor can be determined:ε =NAtcal(2.29)If the reconstructed image is designated by Irec, and the duration of each pro-jection to obtain this image is tp, then one can obtain another 3D image in units ofactivity IA byIA =1εIrectp(2.30)The term Irec/tp represents the measured count rate in each voxel. If deadtimecorrections are required, then a deadtime correction factor δ should be included472.6. Closing Remarksin equation 2.30IA =1εIrectpδ (2.31)in order to obtain a quantitative 3D distribution of activity.A more detailed discussion regarding the measurement of ε and the calculationof δ are given in Chapters 4 and 5 respectively.2.6 Closing RemarksThis chapter briefly summarized the procedure for quantitative imaging. It dis-cussed nuclear decay and its different modes, interactions of radiation with mat-ter relevant to the 177Lu decay, how images are obtained from a gamma camera,and finally it introduced the concept of internal dosimetry. In order to accuratelymeasure dose absorbed by a patient, the temporal behavior of activity within thepatient needs to be determined. By applying corrections for attenuation, scatter,and deadtime it is possible to accurately reconstruct a quantitative 3D image ofthe activity distribution. If this procedure is performed several times at differentdays after injection, then it is possible to obtain the behavior of change in activitydistribution with respect to time. Once this is known, the cumulative activity A˜can be determined.48Chapter 3Bremsstrahlung Characteristics of177Lu3.1 IntroductionTargeted radionuclide therapy (TRT) uses radiopharmaceuticals that emit particlesto deliver dose directly to tumor sites [10]. To verify radiopharmaceutical activitydistributions and to estimate doses delivered to tumors and healthy organs, quan-titative imaging is required [60]. Although beta (β ) particles that are used in themajority of TRT have high ‘killing power’ they cannot be directly imaged. On theother hand, their interactions with tissue result in the emission of BRS radiationwhich can be detected by a gamma camera.With the development of solid state detectors in the second half of the pastcentury, high resolution investigations of BRS spectra became possible. In par-ticular, experimental tests of theoretically calculated BRS production yields havebeen performed [82]. However, in the majority of these studies metallic targets,such as Cu, Cd, Ta, and Pb, have been used [83]. In parallel, synchrotron radiationwith dilute gas targets [84] have also been investigated.493.1. IntroductionUnfortunately, little is known about the production of BRS in human tissue.As the popularity of radionuclide therapies rapidly increases, the BRS effect fol-lowing administration of β emitting radiopharmaceuticals becomes more rele-vant. The information about BRS production may be important for both accurateSPECT/CT quantification and internal dose calculation.Some radioisotopes used in TRT, such as 90Y, are pure β -emitters. Althoughrecently PET imaging of extremely weak 90Y positron emission have been pro-posed [85], traditionally BRS imaging has been considered as the main method tobe used to generate images of distribution of 90Y activity in the body [86]. TheseBRS images, however, have poor resolution and cannot be easily translated intodose estimates, although methods to improve accuracy of 90Y BRS planar [87]and SPECT imaging have been proposed.[88–91] Other radioisotopes availablefor TRT (such as 177Lu, 131I, 188Re), besides β ’s, have gamma emissions that canbe used for imaging. However, due to the presence of β ’s, these gamma emis-sions are always accompanied by a BRS background, which potentially can leadto difficulties in quantification.In order to optimize acquisition parameters in medical imaging studies andobtain accurate quantification of images, it is important to understand the BRSspectrum so that it can be correctly accounted for by the image reconstructionalgorithm. Additionally, without this knowledge, the optimal acquisition param-eters (i.e. energy windows and collimator) may in some cases be difficult to de-termine. In the past, MC simulation studies have been performed investigating90Y BRS, aiming to optimize its acquisition parameters [90, 92, 93] using a HE503.1. Introductioncollimator. Similar studies testing imaging conditions have also been done with131I [94]. Moreover, MC simulations were used to analyze primary and scatteredcomponents in 177Lu ,131I and 90Y projection images [95]. However, to the bestof our knowledge, no detailed studies of BRS production due to 177Lu have beenperformed.Therefore, the purpose of the current study is to expand the scope of theseinvestigations. To this end, the characteristics of BRS spectra produced in tissueby 177Lu was investigated. This was done to examine the effects of BRS on med-ical imaging studies, and to provide guidance for camera setup that would leadto improved quality and quantitative accuracy of images. BRS production andthe spectra detected by a gamma camera were studied using MC simulations. Asour simulations were to be compared with experimental data, we filled both simu-lated and physical phantoms with water since it is an easily available medium thataccurately approximates photon interactions with tissue.There are several MC codes that could be used for BRS simulations. Althoughsome discrepancies between 90Y, 131I, and 188Re voxel S-values determined usingthree different codes (MCNP4C, EGSnrc, and GEANT4) have been found [96],a general conclusion of this study was that these differences will have negligibleimpact on dose calculations. Similarly, small differences in 90Y BRS abundancessimulated with GATE [97] and MCNP were identified [90]. On the other hand,Autret at al. [94] showed good agreement between 131I spectra of point sourcessimulated using MCNP and GATE, and those acquired experimentally. To checkif similar differences, if any, exist when simulating isotopes other than 90Y, the513.2. Materials and MethodsBRS spectra generated using two of these MC codes (namely, Monte Carlo N-Particle V.5 (MCNP5) and Geant4 application for tomographic emission (GATE))were compared.First, the two codes were used to investigate the dependence of BRS produc-tion yields on electron energy using sources emitting mono-energetic electronsplaced in water. Subsequently, a 177Lu source placed in a water-filled cylinderwas simulated. The BRS total yields, the yields for photons with energies higherthan 50keV, the shape of spectra, and their mean energies were determined. Fi-nally, the spectra that would be acquired using a typical SPECT camera (SymbiaT,Siemens Medical, Germany) with LEHR, Medium Energy (MELP), and High En-ergy (HE) collimators were simulated and compared with those obtained exper-imentally. These simulations allowed us to identify the different components ofthe detected energy spectra. A detailed analysis of these data was subsequentlyused to suggest acquisition parameters (collimators and energy windows) whichwould result in optimized imaging conditions.3.2 Materials and MethodsThe study was divided into three parts. First, the BRS spectra generated by GATEand by MCNP5 codes were compared. To do this, simulations of mono-energeticelectrons, as well as 177Lu sources were performed using the two codes. Thedetails of the 177Lu decay are summarized in Table 2.2. The second part of thestudy involved GATE simulations of a planar acquisition using a gamma camera.523.2. Materials and MethodsA small sphere filled with uniform activity of 177Lu was placed inside a cylinderfilled with water. The spectra of all photons which werea) generated inside the phantom.b) photons which were not absorbed and exited the phantom.c) photons which were detected by the camera.were analyzed and compared. The third part, done in parallel with the simulations,involved experiments to measure the spectra with a clinical SPECT camera usingthe same conditions as specified in our simulations.3.2.1 Simulations of BRS using GATE and MCNP5The BRS spectrum can be considered as being a convolution of the continuousspectrum of beta emissions with the continuous spectrum of BRS effect producedat each electron energy. To isolate the dependence of BRS production on elec-tron energy, simulations of point sources of mono-energetic electrons placed inthe spheres filled with water (r = 3cm) were performed using both MC codes:MCNP5 [98] and GATE V.6.1 [99]. The energies of the simulated electrons were100keV , 500keV , 1MeV and 2MeV . This test investigated if there are any differ-ences in BRS creation by these two codes.A second set of simulations involved modeling the same spheres with a pointsource of 177Lu. GATE and MCNP5 use different methods to generate the βspectra. MCNP5 requires the user to provide the β spectrum of the investigated533.2. Materials and Methodsradioisotope. In our case, nuclear decay data from the RAdiation Dose Assess-ment Resource (RADAR) [100] was used as input. GATE uses the evaluatednuclear structure data file (ENSDF) database [101] from Generation and track-ing V.4 (GEANT4) [102] to generate the spectra internally. Simulations of β -decaying isotopes would show the combined effects of BRS production and βspectrum creation by GATE and MCNP5.The shapes of the BRS energy spectra produced by the two codes were an-alyzed. The Total Bremsstrahlung Yield (TBY), defined as the ratio of photonscreated by the BRS process (γBremss) relative to the total number of primary elec-trons (Ne), was evaluated using the following equation:TBY =γBremssNe(3.1)177Lu decays 100% through β− emissions, therefore Ne’s in these case wasequal to the number of decays. While only primary electrons were considered asNe, γBremss included not only BRS photons generated by the primary electrons, butalso those generated by electrons created in secondary processes, such as ioniza-tion and internal conversion.In addition, as the lowest energy which could be detected by SPECT camerasis about 50keV, the BRS yield for photons with energy higher than 50 keV (BY50)was calculated:BY50 =γBremss>50keVNe(3.2)543.2. Materials and Methodswhere γBremss>50keV are photons generated by the BRS process whose energy ishigher than 50keV . The mean Bremsstrahlung energy (MBE) was determinedfrom the data obtained from these simulations. In all cases the simulations wereperformed for Ne = 5×107 electrons.The percent differences in the BRS yield (BYdi f f ) generated by GATE (BYG)and MCNP5 (BYM) were evaluated for mono-energetic electrons and 177Lu usingthe following equation:BYdi f f =BYM−BYGBYG×100 (3.3)3.2.2 Simulation of spectra detected by SPECT cameraCamera ModelingComparison of the results of simulations performed using MCNP5 and GATEshowed little difference between the BRS yields and energy spectra created bythese two programs. On the other hand, GATE has a clear advantage over MCNP5when performing simulations of PET and SPECT acquisitions because it providessubroutines that facilitate camera modeling. A pre-built SPECT system availablein GATE includes a digitizer that accounts for the detector response and the signalprocessing chain. For these reasons, GATE was used in the subsequent simula-tions.A SymbiaT (Siemens Healthcare, Erlangen, Germany) SPECT camera withLEHR, MELP, and HE collimators was modeled. The characteristics of the col-553.2. Materials and MethodsCollimatorTypeHoleDiameter[cm]SeptalLength[cm]SeptalThickness[cm]Sensitivity@10cm[cpm/µCi]GeometricResolution@10cm [mm]SystemResolution@10cm [mm]LEHR 0.111 2.405 0.016 202 6.4 7.5MELP 0.294 4.064 0.114 275 10.8 12.5HE 0.400 5.970 0.200 135 13.2 13.4Table 3.1: Characteristics of the SymbiaT collimators which were used in the sim-ulations and experimental acquisitions discussed in this work (data from Siemens[2]).limators were obtained from the data-sheet publicly available online [2], and areshown in Table 3.1. The detector was simulated as a 3/8” thick NaI crystal. Itwas covered by a 0.5mm thick aluminum layer [103] at the front, while the backcompartment of the camera (photomultipliers) was modeled as a 0.95 cm thicklightguide made of glass followed by 5.65 cm of a material composed of 23%glass, 56% vacuum, and 21% air [104].The camera head shielding was modeled as a lead layer with a thickness of4cm at the sides and 3cm at the back [105]. The energy resolution was specifiedin GATE with a full width half maximum (FWHM) of 10% at a reference energyof 140keV. The dependence of resolution on energy was modeled internally byGATE using an inverse square root law (R∼ 1/√E).Phantom SimulationsThree simulation runs, modeling experimental acquisitions, were performed (threecollimators). The simulation experiments used uniform activity of 177Lu, dis-solved in water placed inside a hollow sphere made of plastic (ρplastic= 1.18g/cm3),563.2. Materials and Methodsas to reproduce the experimental conditions. The internal radius of the sphere wasrs = 0.9cm. The thickness of the plastic shell was tshell = 0.1cm, and the distancefrom the center of the sphere to the collimator surface was set to be Rcam = 25cm.The sphere was placed at the center of a cylinder (r = 11.1cm, h= 19.5cm) filledwith water.The analysis of these simulation results was performed at the three successivelevels analyzing the following characteristics:1. The total number of photons generated inside the water cylinder .2. The total number of photons that were not absorbed and escaped the phan-tom.3. The energy spectra of photons detected by the camera for the three collima-tors.The first two levels involved calculating the TBY (eq. 3.1) and total non-BRSyield (NBY), where NBY is defined as the ratio of photons produced by all pro-cesses other than BRS (γnon−BRS) to Ne:NBY =γnon−BRSNe(3.4)In all cases, the mean energy of the BRS photons was also calculated.The difference of the total number of photons recorded at levels 1 and 2 reflectsthe effect of photon attenuation in water.At the third level, the energy spectra recorded by the camera were simulated573.2. Materials and Methodsfor all three collimators. In this case, the detected photons were separated into thefollowing components:• Primary Gammas: γ-photons generated by the radioactive decay of thesource that were not attenuated and that did not scatter before being de-tected.• Scatter Gammas: γ-photons generated by the radioactive decay of thesource that were not attenuated but scattered before being detected.• Primary BRS: photons generated by the BRS process by β ’s emitted bythe source that were not attenuated and that did not scatter.• BRS scattered: photons generated by the BRS process by β ’s emitted bythe source that were not attenuated but did scatter. Scattering in the phantomand in all elements of the camera head and collimator were included.• Secondary BRS: photons generated by the BRS process caused by sec-ondary electrons.• Other (non-BRS or primary): photons generated by any process otherthan radioactive decay or BRS. These were mostly X-rays generated byinteractions in the collimator.GATE includes the so called actors or sets of tools that allow the user to interactwith the simulation and collect different types of information. The phase spaceactor was used to compare the production of photons inside the phantom with583.2. Materials and Methodsthose that exit the phantom (levels 1 and 2). For level three, a new actor wascoded to correctly identify scattered photons, in particular those that scattered inthe crystal and deposited only part of their energy, and to identify the processwhich generated them (e.g. radioactive vs. BRS).For the first two levels of analysis, 20×106 decays were simulated. For levelthree, 3×109 decays were simulated. The simulations were performed using a 64core computer with 128GB of RAM, and all the cores were used at the same timeusing parallel computing.The MBE was calculated for BRS photons that were generated (MBEgen) inthe phantom, as well as for the BRS photons that were not attenuated and escapedthe phantom (MBEout). The fraction of photons that were absorbed (%γabs) wascalculated. It was defined as the percentage difference between the total numberof photons generated in the phantom (PYgen) and those photons that escaped thephantom(PYout).%γabs =PYgen−PYoutPYgen×100 (3.5)Finally, considering only photons which escaped the phantom, we calculatedthe ratio κ of BRS photons (γBout) to the total number of photons (PYout).κ =γBoutPYout×100 (3.6)593.3. Results3.2.3 Phantom ExperimentsIn order to validate the simulations, three phantom experiments were performedat the Nuclear Medicine department of the Vancouver General Hospital (Vancou-ver, Canada) using the same SymbiaT SPECT camera which was modeled in oursimulations. The experimental conditions (i.e. phantom sizes, acquisitions ge-ometry, collimators, etc) were identical to those used in the simulations (section3.2.2). The 177Lu activity in the small sphere in these experiments was equal to32.8MBq.The energy spectra were collected for three collimators, namely LEHR, MELP,and HE. ASCII files containing the measured energy spectra were exported di-rectly from the manufacturer’s computer and rescaled to match the correspondingsimulated spectra. The normalization used the same approach as proposed byHeard et al [92] where the sums of counts in the energy interval from 100keV to500keV were matched.3.3 Results3.3.1 Simulations of BRS using GATE and MCNP5Figures 3.1a and 3.1b compare the energy spectra of BRS photons generated bythe two codes, MCNP5 and GATE, for mono-energetic electrons and 177Lu re-spectively, and Table 3.2 summarizes their mean energies. These MBE’s wereestimated for the entire spectrum of BRS photons and, additionally, for photons603.3. ResultsEnergy[keV]MBE [keV] MBE>50 [keV]GATE MCNP5 GATE MCNP5100 6.6 8.8 63.0 62.8500 18.8 22.4 113.8 112.61000 31.6 35.7 155.4 152.72000 53.7 60.7 224.0 221.1177Lu 10.4 13.4 88.3 87.9Table 3.2: Mean energy of Bremsstrahlung photons (mean Bremsstrahlung energy- MBE) created by mono-energetic electrons and β decay of 177Lu.with energies higher than 50keV . Figure 3.2 shows the BRS yields, namely TBYand BY50, obtained from the simulations using GATE and MCNP for differentelectron energies and isotopes<1E-3<1E-30.0100.0070.0060.0060.0530.0480.0180.0190.1100.1090.0540.0600.2570.252BY50 GATE BY50 MCNP TBY GATE TBY MCNP0.000.050.100.150.200.25Bremsstrahlung Yield 100 keV  500 keV  1000 keV  2000 keV(a) Mono-energetic electrons<1E-3<1E-30.010.01BY-50 GATE BY-50 MCNP TBY GATE TBY MCNP0.0000.0020.0040.0060.0080.0100.0120.014Bremsstrahlung Yield 177Lu(b) RadioisotopesFigure 3.2: Comparison of total Bremsstrahlung yield (TBY) and Bremsstrahlungyield for photons with energies higher than 50keV (BY50) obtained from simula-tions using GATE and MCNP5. The yields of Bremsstrahlung photons generatedby mono-energetic electrons (a) and by the β emissions from the radioisotopes (b)are shown.613.3. ResultsEnergy [keV]0 200 400 600 800 1000 1200 1400 1600 1800 2000Bremsstrahlung Yield-710-610-510-410-310-210Bremsstrahlung Yield vs. Energy100 keV500 keV1000 keV2000 keV(a)(b)Figure 3.1: Bremsstrahlung spectra generated by (a) mono-energetic electrons andby (b) the β emissions from 177Lu generated with GATE (solid lines) and MCNP5(dashed lines). The spectra were normalized to the total number of primary elec-trons.623.3. ResultsFigure 3.3: Total yields of Bremsstrahlung (TBY) and non-Bremsstrahlung (NBY) photons generated by β ’s emitted by 177Lu, in thewater-filled cylinder and the corresponding total yields of photons that escapedthe phantom.3.3.2 Comparison of Simulations and Phantom ExperimentsFigure 3.3 presents the BRS and non-BRS yields calculated using GATE simula-tions of the source inside the water-filled cylinder, and for the photons that escapedthe phantom without being attenuated, respectively.Figure 3.4a compares the measured and simulated energy spectra for threecollimators of the sphere containing activity placed in the water-filled cylinder.Figure 3.4b shows the components of the simulated spectra presented in Fig-633.3. Resultsure 3.4a. The results of quantitative analysis of these simulations show that theMBEgen = 10.4 keV, MBEout = 64.2 keV, %γabs = 34%, and κwater = 0.3%.(a) Measured vs. Simulations. (b) Simulated components.Figure 3.4: Analysis of energy spectra of photons recorded by the Siemens Sym-biaT SPECT camera and those obtained from GATE simulationsFigure 3.5 shows the relative composition of the spectra recorded by the cam-era, for each collimator.643.3. ResultsFigure 3.5: Percentage contributions of each type of photons to the total numberof detected photons for different collimators.653.4. Discussion3.4 Discussion3.4.1 Bremsstrahlung Spectra Simulated using GATE andMCNP5 programsThe results of our simulations indicate that for 100keV mono-energetic electrons,BYG are higher than BYM by 27% , but only by 11% for 500keV electrons (seeFig. 3.2a). For higher electron energies, this difference decreases to 1% and 2%for the 1MeV and 2MeV electrons, respectively. When comparing BY50, for the100keV and 500keV electrons this difference substantially decreases, while forhigher electron energies the trend reverses and MCNP5 exceeds GATE by about11% for both 1MeV and 2MeV . Similarly, when analyzing BRS production from177Lu β emissions, BYG is higher (by 27%) than BYM (see Fig. 3.2b). However, forBY50, these differences decrease, becoming less than 10%. These results agreewith those of Rong et al. [90] who noticed similar differences between 90Y BRSspectra simulated by MCNP and GATE. In general, MCNP5 generated fewer BRSphotons, but for energies relevant to nuclear medicine imaging (E ≥ 50keV ) thedifferences in BRS generation are small. Figure 3.1 compares the BRS energyspectra obtained from both MC codes. In summary, the biggest differences inBRS production occur at low energies, with GATE producing higher yield.The differences in BRS production by the two MC codes are also reflectedin the corresponding MBE values (Table 3.2 ). The values of MBE from MCNP5are always higher than those obtained with GATE, with the differences of up to33% for the case of the mono-energetic electrons and 22% for 177Lu. However,663.4. Discussionwhen only photons with energies higher than 50keV are taken into account, MBEGATE values are higher and differences decrease to -1.7% for the mono-energeticelectrons and less than 0.5% for 177Lu. These results confirm that both codes havevery similar behavior in the energy range relevant for nuclear medicine imagingstudies (i.e photon energies higher than 50keV), which justifies our decision touse GATE in all subsequent simulations presented in this work.3.4.2 Comparison of Simulations and Phantom ExperimentsDirect quantitative comparison of the simulated and experimental spectra is dif-ficult because normalization, which must be applied here, is rather arbitrary. Weopted to follow a similar normalization procedure as Heard et al. [92] in whichthe areas under the curves between 100keV and 500keV were matched. Regard-ing the camera modeling, our simulation was approximate and did not reflect thedetails of the geometry of our experimental studies.In spite of these limitations, the shapes of the spectra obtained from simula-tions reproduce relatively well those from experimental measurements (Fig. 3.4a).The differences seen in all spectra at low photon energies, where the simulationsoverestimate the measured spectra, are due to the fact that most SPECT cam-eras have a cutoff energy below which no photons are accepted which is typicallyset around 50keV, while no cutoff was applied to simulated spectra. Anotherarea where experiments and simulations differ occurs at photon energies around100-120keV. There, in almost all cases simulations exceed experimental data sug-gesting increased amount of scattered photons in the simulations. This effect is673.4. Discussionprobably due to discrepancies in the modeled camera geometry.A detailed examination of the relationship between the accuracy of cameramodeling (with medium energy general purpose (MEGP) collimator) and the shapesof the energy spectra simulated by GATE was performed by Rault et al.[104] Thecamera model used in our simulation would correspond to the Intermediate Modelused in the Rault’s study. Although direct comparison of the simulations results isimpossible because 177Lu was not included in Rault’s study, the general trends arevery similar. In particular, similar excess of scattered photons in the 100-120keVregion as seen in our study, was also observed in 131I and 18F spectra simulatedby Rault. The normalization seem to overestimate the 208keV peak for simulatedspectra acquired with MELP and HE collimator.In general however, the shapes of the simulated spectra reproduce well thoseobtained from experiments, which gives us confidence that the analysis of thedifferent spectra components (section 3.4.4), provides correct information abouttheir relative contributions to the total spectrum.3.4.3 Quantitative Analysis of Bremsstrahlung SimulationsAt the next stage, the numbers of photons produced inside the phantom, and thosethat escaped the phantom (Fig. 3.3 ) were evaluated and the following character-istics were observed:• The maximum β ranges for 177Lu (see Table 2.2) are small compared withthe sphere’s dimensions (the sphere diameter was 1.8cm). However, as683.4. Discussionthe spheres contained uniform isotope solution, some electrons which wereemitted close to the sphere surface may have escaped the sphere and in-teracted with the water inside the cylinder. Comparable TBY were ob-tained for 177Lu and the mono-energetic 100keV electrons (Figs. 3.2 and3.3) which again could be expected considering the mean energy of the177Lu decay. These results may have implication for dosimetry calcula-tions performed using voxel S-value approach, as they show that althoughBRS production in the volume exceeding electron range is small (about10%), nevertheless it contributes to the total dose, and the region where thisradiative BRS energy is deposited extends over several voxels.• Due to the fact that about 95% of BRS photons have energies below 50keV,many of them are absorbed in the phantom. Because attenuation is strongerfor low energy photons than for those with high energies, MBE of BRSphotons that escaped the phantom relative to MBE of generated photons isshifted towards higher energies (similar to “beam hardening” of X-rays). Inthis case the increase was from 10.4keV to 64.2keV.• Attenuation affects BRS photons to a higher degree than photons from γ-decay due to their lower energy. As it can be seen in Figure 3.2b, more than99.9% of BRS photons have energies below 50keV, where absorption insidethe phantom plays a big role.• The fractions of BRS photons that leave the phantom relative to the totalnumber of generated photons account for only 0.3%.693.4. Discussion3.4.4 Spectra Analysis and Optimization of AcquisitionParametersSimulations provide an excellent opportunity to isolate groups of photons thatwere created through different processes and have or have not scattered in themedia, and to analyze their relative contributions to the spectra (see Figure 3.4b).Such analysis may be very helpful in identifying acquisition parameters (colli-mators and energy windows) which would lead to best quality images becauseunscattered photons carry most of information about source position.This type of analysis performed by Heard et al. [92] in simulations of 90Yfor low and medium energy collimators. To the best of our knowledge, no suchanalysis was done for 177Lu.For 177Lu the number of BRS photons is much lower than that of other pho-tons. BRS photons (primary and scattered) never exceed 0.2% of total detectedphotons. Therefore, since BRS contribution to the spectrum is very small, the se-lection of collimator and energy windows can be based solely on the analysis ofthe isotope’s gamma emissions. The following summarizes both current practiceand our recommendations.177Lu has two photopeaks, at 113keV and 208keV, that can be used in imag-ing. If only LEHR collimator is available, it has been shown that accurate imagequantification can be achieved [106], but in this case only the 113keV photopeakmust be used as for 208keV photopeak septal penetration is strong which makesimaging very difficult if not impossible. Additionally, the imaging method must703.5. Conclusionsaccount for high background of scatter photons underneath the 113keV photo-peak. Although the 208keV peak dominates the spectrum acquired with LEHRcollimator and the ratio of scatter to primary photons for this peak is lower, sub-stantial septal penetration from high energy photons (208, 249, and 321 keV)would make imaging using this peak very difficult (see Figure 3.4b). Therefore,the medium energy collimator is recommended. The differences in spectral con-tents measured with MELP and HE collimators are minimal, but the former hasbetter sensitivity and resolution. Medium energy collimators have been used inthe majority of clinical imaging studies of 177Lu [107]. Since simulations showthat BRS contributes less than 0.2% to the detected energy spectrum, no separatecorrections for this effect are necessary.Finally, our simulations indicated that secondary BRS production was alwaysbelow 0.1% of the total detected photons. Lead X-rays due to photon interac-tions with the collimator were observed, with a contribution of less than 4% formedium and high energy collimators. However, Pb X-rays have well known en-ergies (which are usually below the imaging range), therefore they can be easilyavoided when imaging.3.5 ConclusionsThe aim of this study was to investigate the characteristics of BRS radiation pro-duced in tissue by 177Lu, which is a β -emitting radioisotopes used in TRT. Thiswas done in order to improve our understanding of the energy spectra acquired713.5. Conclusionsduring medical imaging studies as this would allow us to identify camera configu-ration which would result in best image quality and apply the appropriate correc-tions to obtain good quantitative accuracy of images.The simulations showed not only that the BRS yields are very low, but alsothat most of the created photons have energies below 50keV. In the 0-50keV en-ergy range, the BRS contribution to the spectra recorded by the camera would beminimal due to strong absorption of low energy photons in the phantom (or in thepatient) and low energy discriminator setting of the camera. For this reason, whenimaging 177Lu, BRS contributions to the energy spectra detected by the camerawould be less than 0.2%. However, background of scattered high energy photonsmust be corrected for when quantitative activity determination is required. Thisbackground is especially pronounced in studies performed with LEHR collimatorthat require a photopeak window around the 113 keV peak.The analysis of the 177Lu simulated spectra indicates that optimal imagingconditions for this radioisotope will be achieved when using MELP collimators.This will allow imaging of the 208 keV photopeak without star artifacts generatedby septal penetration. Moreover, the contribution of scatter under this photopeakis less than in the 113keV.Our results suggest then that Bremsstrahlung contributions to the detected en-ergy spectrum in imaging studies of 177Lu have no degrading effects in imagequantification. We recommend the use of medium energy collimators for whichan energy window located in the 208 keV photopeak with possible windows oneach of the sides to correct for scatter if a method like the TEW is employed.72Chapter 4Quantification of 177Lu in Phantoms4.1 IntroductionBesides chemotherapy and external beam radiation therapy, PRRT using 177Lu hasbeen recognized as an effective tool to treat NETs [38, 55, 108–111]. These tu-mors overexpress somatostatin receptors which can be targeted with radiolabeledanalogue peptides such as DOTATATE, DOTANOC, and DOTATOC to preferen-tially kill tumor cells.Lutetium-177 is a β decaying isotope with a 6.6 days half-life. Beside β ′s, italso emits several gamma rays. The two most abundant lines occur at 113 keVand 208 keV with intensities of 6.2% and 10.3%, respectively [65]. This typeof emissions makes 177Lu very useful for PRRT because the β particles deliverenergy to kill tumor cells, while the gammas can be used in SPECT studies totrack distribution of the radiopharmaceutical within the patient.The current limitations of this treatment is that all patients are injected withthe same activity of approximately 7400 MBq/cycle [112] with treatment per-formed in four cycles. This activity has been determined based on renal toxicitylevels from external beam radiotherapy studies, aiming to not exceed a maximum734.1. Introductionkidney dose of 23[42] or 27 Gy [57]. However, radiotracer uptake in tumors andhealthy tissue varies greatly between patients, leaving a big group of under-treatedindividuals [20, 113]. It is generally believed that treatment plans using injectionsbased on an individualized dose assessment could significantly improve PRRToutcomes and such personalized approach should become routine as it is in exter-nal beam therapies [58, 114].In order to obtain an accurate estimate of the dose, quantitatively accurate de-termination of the activity distribution within the patient and information abouttemporal changes of this activity are required. The temporal behavior is deter-mined from a series of nuclear medicine scans usually performed during a weekfollowing the injection. The distribution of activity requires that corrections forphysical processes that degrade image quality are applied when performing im-age reconstruction [60]. Additionally, in order to convert image counts to activityvalues, the efficiency of the camera (calibration factor) needs to be measured.Different groups have studied ways to determine the camera calibration fac-tor. Dewaraja et al. [115] considered three different geometries in an 131I study.They used a point source, a uniform filled phantom, and a hot sphere centered ina phantom with background activity. They concluded that using the calibrationobtained from the hot sphere placed inside a large phantom was superior to theone obtained from the uniform phantom, but similar to the point source methodonce a certain size for the volume of interest (VOI) around the source in the pla-nar scan was reached. Ljungberg et al. [116] also performed a study with 131I.They measured the calibration factor using a planar scan of a point source and744.1. Introductionalso performed simulations. Counts in the whole field of view (FOV) were takeninto account, and a good agreement was obtained between simulations and mea-surement. Zeintl et al. [117] used a phantom uniformly filled with 99mTc and theyreported similar quantification accuracies as previously obtained by our group us-ing planar scans of point sources [118]. Frey et al. [119] suggests that the use ofa phantom tomographic acquisition, instead of a planar scan of a small source, isthe optimal approach.In two more recent studies, Anizan et al. [120] have been trying to test therepeatability and stability over time of the camera sensitivity determination. Theymention that the largest source of variability comes from the measurement of theactivity in the dose calibrator, and source preparation. Background activity is thesecond factor influencing variability in the measurement, but camera sensitivitywas seen to be very stable over time. They however, did not compare the resultswith a SPECT acquisition. The second study [121] investigated the used of planarscans of sealed sources with long half lives to avoid the source preparation error.They concluded that calibration factors for cameras of different manufacturersare different, and that separate calibrations should be performed for each camera.Also, it is mentioned that an ROI around the source should be carefully selectedin order to prevent contamination with background counts. The ROI was selectedbased on the extent of the object in the projection plane.Based on these studies, one can conclude that the determination of the cam-era calibration factor is not standard. Planar and SPECT acquisitions have beenperformed with different geometries, reconstructed with different corrections, and754.1. Introductionusing different radionuclides. However, the accurate determination of the calibra-tion factor is necessary for image quantification.The next step is concerned with studies focused on specifying an imaging pro-tocol that can easily be implemented clinically. Beauregard et al. [107] used cam-era manufacturer’s software which applies attenuation and TEW scatter correc-tions to obtain quantitative images of 177Lu. The method was tested with 175mLcylindrical inserts with VOI’s selected using fixed thresholds with values between1-40%.The results of our first 177Lu quantification studies were published by Shcherbininet al. [106] in which it was shown that imaging using the 113keV photopeakwith LEHR collimators can lead to accurate quantification (2% accuracy for a70mL container scanned in air and water without background activity). The APDImethod for scatter correction, based on analytical determination of scattered pho-tons using the Klein-Nishina formula, was used for scatter correction.Seven scatter correction methods were studied by de Nijs et al. [122] focusingon 177Lu quantification with different collimators. They analyzed six spheres withvolumes ranging from 0.5-27mL, placed in a uniform background with a ratio ofspheres to background activities of SBR≈ 13. They mention that TEW might notbe very suitable when using data obtained from the 113keV gamma peak. How-ever, for the 208keV peak, they suggest that the TEW method could be changedinto a dual window method because the higher window contains mostly noise thataffects the accuracy of the method. Their upper scatter window (USW) had ahigher limit of 239.2keV compared to our 280.2keV one.764.1. IntroductionSanders et al. [123] performed quantification for a cylinder uniformly filledwith activity and for spheres with volumes ranging from 0.5mL to 16mL usingTEW scatter correction. Their results provided quantification accuracies of 20%or greater with the error mostly due to partial volume effects. All the work pub-lished to date suggest that superior quantification is achieved when reconstructionsare performed with data obtained from the 208keV peak rather than the 113keV.The aim of the current study was to develop a method to accurately determineactivity quantification of 177Lu in the simplest possible way. Two different scattercorrection methods, APDI [80, 81] and TEW [79], planar vs. tomographic de-termination of the camera normalization factor, and three different segmentationmethods were investigated.APDI is an analytical scatter correction method that uses information aboutactivity distribution in the object from a first OSEM reconstruction and the attenu-ation map of the object and the Klein-Nishina formula to estimate the contributionof scattered photons to the measured projections. The method has been shown toprovide accurate quantification for 99mTc, 188Re, and 177Lu [81, 118, 124, 125].The limitation of APDI is that it is very computer intensive. TEW on the otherhand, is fast and it has been implemented on many SPECT/CT cameras making itavailable to every Nuclear Medicine department, however, it does not model thephysics of scatter and is just an approximation.Regarding camera normalization, planar acquisitions are easier and faster toperform than tomographic scans but it is believed that they can only be used whenhighly accurate scatter and attenuation corrections during the tomographic image774.2. Materials and Methodsreconstruction are performed. Tomographic acquisitions are said to better ap-proximate scatter and attenuation in patients resulting in better compensation forpossible errors in the quantitative reconstructions. [60]Finally, in order to perform organ dosimetry exact information about organactivity and mass is required. For this, we compared segmentation using a fixedthreshold of 40% which is commonly used in clinical environments [126], seg-mentation based on true organ volume obtained directly from CT, and an itera-tive segmentation method based on calibration curves for different signal to back-ground ratio (SBR)’s [127].4.2 Materials and MethodsIn order to obtain the accurate activity in an ROI several steps are required:1. Data acquisition following and optimized protocol.2. Image reconstruction with attenuation and scatter corrections to obtain areconstructed 3D image in which the value of each voxel represents thenumber of photons detected in that particular location.3. The determination of the camera normalization factor in order to convertthe photon counts into activity values (e.g. MBq).4. If organ dosimetry is desired, the final step is to accurately segment theorgans.784.2. Materials and MethodsFigure 4.1: Work flow for the generation of quantitative 177Lu images. Imagereconstruction with attenuation and scatter correction, in combination with a cor-rectly determined camera normalization factor generate an image containing theactivity distribution. Segmentation for the determination of the dose in ROI’s isperformed once the activity distribution has been determined.794.2. Materials and MethodsFigure 4.1 shows a diagram of these steps.To test our work flow and determine the accuracy of quantification we per-formed a series of phantom experiments under different attenuation and scatterconditions:1. Phantoms in air: The purpose of scanning hot objects of different shapesand sizes placed in air was to evaluate the accuracy of our reconstructionmethod with minimal amount of attenuation and scatter.2. Phantoms in water: In this case hot objects were placed in cylinders filledwith water and allowed us to evaluate the effects of our attenuation andscatter corrections on quantitative accuracy of 177Lu. It also allowed us tocompare the results of the two scatter correction methods; APDI and TEW.3. Phantoms in hot water: The purpose of these scans was to test the segmenta-tion algorithms once the accuracy of our quantitative method was evaluatedwith the phantoms in air and water. This situation approximates well theconditions present in patient studies.4. We designed a challenging situation in which we placed four bottles withactivity on the camera bed so that the objects covered the total area of thedetector. We also placed these bottles between water bags to generate addi-tional attenuation and scatter.The work flow details are explained in the upcoming sections and Table 4.3 sum-marizes the information about phantom configurations, volumes, and activities ofinserts and acquisition times per projection.804.2. Materials and MethodsName Lower Limit [keV] Center [keV] Upper Limit [keV]LSW 153.0 170.0 187.0PW 187.2 208.0 228.8USW 229.5 255.0 280.2Table 4.1: Energy window setup used for phantom experiments of 177Lu .4.2.1 Data AcquisitionAll scans were performed using a SymbiaT (Siemens Medical, Germany) SPECT/CTcamera with a MELP collimator and non-circular orbit. For each scan 90 projec-tions were acquired using a 128x128 matrix and three energy windows: the 208keV photopeak window (PW), low scatter window (LSW), and upper scatter win-dow (USW) (see Table 4.1). The duration of each projection are listed in Table4.3.4.2.2 Image reconstructionReconstructions of the data were performed using our graphical user interface(GUI) called SPEQToR. This GUI allows the user to select both the reconstruc-tion algorithm and the set of corrections to be included in the reconstruction. Allour reconstructions were performed with the standard OSEM algorithm (using10 subsets and 6 iterations) with attenuation correction (AC), resolution recov-ery (RR), and scatter correction (SC). Two different SC methods were tested:TEW[79] and our APDI [80, 81] algorithm.In the case of 177Lu imaging studies, the scattered photons recorded in the 208keV photopeak window have two components: the self-scatter component and the814.2. Materials and Methodshigh-energy scatter components. Both components have been incorporated intothe denominator of the OSEM formula:X l+1j =X lj∑i∈ΩiCi j∑i∈ΩiCi jYi∑kCikX lk+S+H(4.1)In this equation, Yi represents the measured projections, C is the system matrix inwhich attenuation and resolution recovery information are included, and X is theestimated image. The terms S and H in the denominator represent the self-scatterand high-energy scatter components, respectively.The term S+H is the scatter component calculated using either TEW or APDI.4.2.3 Determination of dose calibrator settingsA vial containing 0.129mL of 177LuCl3 with manufacturer specified activity of10.4± 1.04GBq was used. To perform calibration (determine the dial settings)of our CRC-25R (Capintec, USA) dose calibrator. The activities for both thedetermination of the camera normalization factor and for phantom experimentswere measured using this dial setting.4.2.4 Determination of the Camera Normalization FactorPoint sources and phantoms filled with 177Lu and 99mTc were used to test theaccuracy of camera normalization factor (camera efficiency). Our goal was tocompare normalization factors obtained for an isotope with a single photopeak(99mTc) with that which includes contamination from higher energy emissions824.2. Materials and Methods(177Lu). Table 4.2 summarizes the experiments performed using point sourcesand phantoms. For the case of 99mTc , a LEHR collimator was used, while for177Lu a MELP was selected. The uniform cylinder filled with 177Lu was scannedat the Center Hospitalier Universitaire de Quebec, Quebec City, Canada, using thesame acquisition parameters as described in section 4.2.1.For the planar scans (A and C in Table 4.2), two methods were tested whendetermining the camera calibration factor (CF):1. Using all counts in the PW: The CF was calculated by dividing the sum ofcounts recorded in the PW (Cpw) by the activity of the source (A) and theduration of the scan (td).CF =Cpw(A× td) (4.2)2. Using the counts in the PW corrected for scatter with the TEW method: Thescattered counts (Cs) were estimated using the counts in the LSW (Cls), theUSW (Cus), and the widths of each window wls, wus, and wpw for the LSW,USW, and PW, respectively, and the following formula:Cs =(Clswls+Cuswus)wpw2(4.3)The total number of primary photons (Cprim) in the PW was then determinedby subtracting Cs from CpwCprim =Cpw−Cs (4.4)834.2. Materials and MethodsIsotope Modality Phantom Source Info Acquisition Parameters99mTcA) Planar(point-like)Activity: 22.4MBqVolume: 0.5 mLScan Duration: 10 minCollimator: LEHRPW: [129.5-150.5]keVLSW: [108.5-129.5]keVUSW: Assumed zeroB) SPECT(NEMA)Total Activity: 689.1MBqVolume of spheres: [0.5-37] mLNumber of Projections: 120Duration per Projection: 30sC) Planar(point-like)Activity: 11.7MBqVolume: 0.5 mLScan Duration: 10 minD) SPECT(Cylinder)Total Activity: 659.6MBqVolume: 6500 mLNumber of Projections: 96Duration per Projection: 10sCollimator: MELPPW: [187.2-228.8]LSW: [153.0-187.0]USW: [229.5-280.2]177LuE) SPECT(Jaszczak)Total Activity: 459.1MBqVolume of Spheres: [0.5-113] mLNumber of Projections: 90Duration per Projection: 20sF) SPECT(Jaszczak HotBackground)Total Activity: 2486.6MBqVolume of Spheres: [0.5-113] mLNumber of Projections: 90Duration per Projection: 30sTable 4.2: Phantoms used in planar and tomographic acquisitions in determinationof the camera normalization factor for both 99mTc and 177Lu.844.2. Materials and MethodsThe CF was calculated using equation 4.2, but using the primary photonsonlyCF =Cprim(A× td) (4.5)For the four tomographic scans (B, D, E and F in Table 4.2), the CF was deter-mined by taking the total number of counts in the reconstructed image (Crec), anddividing them by the activity in the phantom times the total duration of the scan,which was given by the number of projections (np) multiplied by the duration ofeach projection (tp).CF =Crec(A× tp×np) (4.6)The reconstructions of the SPECT/CT data were performed using the methodsdescribed in section 4.2.2, with TEW scatter correction. For the case of 99mTc ,the USW was assumed to have zero counts.4.2.5 SegmentationThe segmentation of the phantom inserts was performed using the following pro-cedure: A large ROI was manually drawn around the desired object. Inside thisregion the following thresholds were used:• For the case of scans in air, a fixed threshold of 0.1% was applied to the datainside the initially drawn large ROI.• For the case of scan in water, a fixed threshold of 1% was applied to the datainside the initially drawn large ROI.854.2. Materials and Methods• In the case of hot background we used three methods:– a fixed threshold of 40% applied to the data inside the initially drawnlarge ROI,– an ROI was created by manually selecting objects based on the CTimages.– Our iterative adaptive dual thresholding (IADT) [127] method.The IADT method aims to determine the volume of the ROI and the ac-tivity contained within it using solely nuclear medicine images (no CT isbeing used). This is done by using two thresholds: one for volume (ThV)and another one for activity (ThA). The values of these thresholds are de-termined in a series of phantom experiments and they depend on the SBRof the segmented object, and due to partial volume effects, the threshold foractivity determination must encompass a larger region than that for the vol-ume determination. Also different reconstruction methods generate differ-ent curves representing dependence of ThV and ThA on SBR. Using thesecurves and a series of iterative steps, the IADT algorithm finds the SBR andthe threshold values appropriate for the investigated object.To determine the values of the thresholds a series of calibration phantomexperiments must be performed. For this task, six bottles were filled withactivity concentration of 0.62±0.03MBq/mL and placed inside a Jaszczakphantom. The volumes of the bottles were 17.0mL, 33.0mL, 199.5mL,864.2. Materials and Methods33.9mL, 76.0mL, and 16.5mL. Activity was gradually added to the back-ground in order to obtain four different SBR (14.0± 1.1,8.0± 0.6,5.3±0.4,4.1± 0.3) and SPECT/CT scans of the were performed. The imageswere reconstructed using OSEM with both TEW and APDI scatter cor-rection methods. The curves obtained for both scatter correction methods(ThV and ThA vs. SBR) were used to segment the experimental data recon-structed using the corresponding scatter correction method.4.2.6 Phantom ExperimentsTo test the quantification accuracy of our methods, eight different phantom con-figurations were used.Jaszczak phantom with spherical insertsSeven spheres with volumes ranging from 0.5mL to 113mL filled with water con-taining 3.19±0.17MBq/mL of 177Lu were placed inside the cylindrical (Jaszczak,Data Spectrum Corporation) phantom (22.2 cm diameter, 19.5 cm height). Thephantom was first scanned with the spheres in air (referred to as Jaszczak spheresin air). Next, the cylinder was filled with water, (referred to as Jaszczak spheresin water), and a second scan was performed.A third scan (referred to as Jaszczak spheres in hot water) was performedwith activity present in the background. The concentration of activity in the back-ground was equal to 0.49±0.03MBq/mL, resulting in SBR of 6.4±0.5. .874.2. Materials and MethodsJaszczak with cylindrical bottle insertsNext, in order to evaluate the potential dependence of the accuracy of quantifi-cation on the shape of the object, we decided to repeat some of the previousscans using cylindrical inserts. Two 8.5mL (B1 and B2), two 11.7mL (B3 andB4), and one 16.2mL (B5) bottles were filled with activity with concentration of5.11±0.22MBq/mL. Similar to the situation with the spherical inserts, the phan-tom was first scanned in water (referred to as Jaszczak Bottles in Water). A secondscan (Jaszczak Bottles in Hot Water) was performed with the phantom filled withwater containing concentration of activity of 0.104±0.003MBq/mL. The SBR inthis case was equal to 49.1±2.6.Thorax phantomThe aim of the third experiment was to determine the accuracy of 177Lu quan-tification in challenging non-uniform attenuation conditions. First, four bottles(T1-T4) of 33.9 mL each, were filled with the activity shown in Table 4.3. Thebottles were attached in different locations inside the Thorax phantom (Ellipticallung-spine body phantom, Data Spectrum Corporation) as follows:• Bottle T1 was attached to the spine.• Bottle T2 was placed under one of the lungs.• Bottle T3 was placed in between the lungs and the spine.• Bottle T4 was attached to a beef bone placed at the bottom of the phantom.884.3. ResultsThis bone was used to model tumor close to an object having true boneattenuation (spine insert only approximates real bone density) and to makethe attenuation at the bottom of the phantom less homogeneous.The phantom was filled with water, sealed, and scanned.Bottles on the camera bedThe final series of experiments aimed to determine the accuracy of quantificationin the most difficult case when bottles containing activity were placed directly onthe camera bed and then between irregularly shaped objects (water bags) creatingnon-uniform attenuation and scatter conditions. Bottles C1, C2, C3, and C4 werefilled with activity and placed on the camera bed. A first scan was performedin air, (Bottles on Bed). Then, the bottles were placed between four 2L waterbags (two below and two on top of the bottles) and a second scan was performed(Bottles on Bed with Water Bags).4.3 Results4.3.1 Determination of the Camera Normalization FactorFigure 4.2 shows the calibration factors obtained with both planar and tomo-graphic scans using 177Lu and 99mTc sources.894.3. ResultsPhantom InsertsPhantomConfigurationProjectionDura-tion[s]NameVolume[mL]Activity[MBq]JaszczakSpheres Air20JaszczakSpheres Water30S0S1S2S3S4S5S60.51.02.04.08.016.0113.11.7±0.13.2±0.26.1±0.312.3±0.624.5±1.248.3±2.4351.6±19.1JaszczakSpheres HotWater30JaszczakBottles Water20B1B2B38.58.511.742.6±2.642.2±2.661.3±3.2JaszczakBottles HotWater20B4B5B611.716.234.060.0±3.290.1±4.390.1±4.3ThoraxPhantom20T1T2T3T433.333.533.834.2300.0±9.9)302.9±10.0302.9±10.0302.9±10.0Bottles on Bed 10C1C233.9163.1205.9±11.1854.0±46.1Bottles on Bedwith WaterBags10C3C4182.1199.1846.5±45.71182.6±63.8Table 4.3: Summary of the different phantom experiments.904.3. Results5.35.055.04Method 1 Method 2 Hot BkgPlanar SPECT0123456Camera Calibration Factor - 99mTc Sensitivity Factor [cpm/kBq](a)0.710.590.60.620.59Method 1 Method 2 Air Cylinder Hot BkgPlanar SPECT0.00.20.40.60.8Sensitivity Factor [cpm/kBq]Camera Calibration Factor - 177Lu (b)Figure 4.2: Camera calibration factors obtained with the two planar acquisitionmethods and the tomographic scans for 99mTc (a) and 177Lu (b).4.3.2 Phantom ExperimentsFigure 4.3a shows the quantification accuracy (percentage difference with respectto the truth) obtained from reconstructions of scans in air. Figure 4.4a shows theaccuracy of quantification for the phantoms scanned in water. The objects havebeen sorted by increasing volumes from left (S1) to right (C4).For the case of scans in air, APDI shows a trend in which all the activitiesare overestimated by up to 11%. In this case, TEW is closer to the true value ofactivity within 7%.For scans performed in water, volumes lower than 2mL present an underesti-mation of activity for both scatter correction methods. For volumes ranging from12mL to 50mL, quantification achieved with APDI underestimates the activitywhile TEW overestimates it, but the values obtained with APDI are closer to thetruth. The bottles within the thorax phantom show similar trends for both meth-ods, with better accuracy provided by APDI for all the bottles in this phantom. In914.3. Resultsparticular, bottle T4, which was attached to a bone, shows remarkable accuracy.Finally, for bottles placed on the camera bed surrounded by the water bags, bothmethods behave very similarly, with very good accuracies for bottles C2-C4.Figures 4.5a, 4.5b, and 4.5c show the accuracy of quantification for the threedifferent segmentation methods when the scans were performed with a hot back-ground. The objects have been sorted by volumes from the smallest object (S2)to the largest one (S6). The 40% threshold underestimates the true activity byup to 50%. When performing the segmentation of the objects using CT basedsegmentation, this again underestimates the activity and the results are similar forboth TEW and APDI. The IADT method, although not perfect, seems to be themost effective in recovering the activity of each object. It is still not very accuratefor small volumes (<17mL) possibly due to partial volume effects. However, forobjects larger than 34mL accuracies better than 10% were achieved.Figures 4.3b, 4.4b, and 4.5d show box plots of the accuracy of quantifica-tion obtained for the different phantom scans. These plots are useful to see thedistribution of quantification accuracy for both APDI and TEW scatter correctionmethods. In patient dosimetry, lesions of different sizes and attenuation conditions(e.g. close to or far away from bones) might be present and our results suggestvariability of the accuracy. It is important to identify possible variations and dis-persion of the accuracy of quantification when comparing the mentioned scatter924.3. Results(a)(b)Figure 4.3: Quantification accuracy for the different phantom inserts scanned inair (a) and its distribution (b) for both scatter correction methods, TEW and APDI.934.3. Results(a)(b)Figure 4.4: Quantification accuracy for the different phantom inserts scanned inwater (a) and its distribution (b) for both scatter correction methods, TEW andAPDI.944.3. Results(a) (b)(c) (d)Figure 4.5: Quantification accuracy for the different phantom inserts scanned inhot water and segmented with three different methods: 40% fixed threshold (a),CT based (b), and IADT (c). The distribution of accuracy is shown in (d) for bothscatter correction algorithms, TEW and APDI.954.4. Discussioncorrections, in order to decide if there are possible advantages of using one com-pared to the other besides quantification accuracy. In order to not overdose thepatient, a conservative approach in which we assume the worst possible quantifi-cation such that it provides us with an upper limit of dose would be the one to betaken into account. Smaller variations around a mean value would suggest a betteroverall performance as long as quantification is within a few percent compared tothe truth.4.4 Discussion4.4.1 Determination of the Camera Normalization FactorThe results of our camera normalization factor are presented in Figure 4.2. Ingeneral for both isotopes, the values of normalization determined using the planarmethod 2 (with scatter correction) and all tomographic scans are very similar.When scatter correction is not applied (planar method 1) the normalization factoris overestimated.• For 99mTc, the planar scan method 2 and the tomographic scan agree towithin 0.2%. However, the calibration factor obtained using the planarmethod 1 still remains within the experimental uncertainty.• The same effect is observed for 177Lu, the planar method 2 and the tomo-graphic scans agree well. Even the calibration factor of the camera in Que-bec City (uniform cylinder) remains within 3%. The planar method 1 results964.4. Discussionin value higher than that obtained from the other methods and the differenceexceeds the experimental uncertainty.The higher values of the calibration factors determined with planar method 1 forboth isotopes are caused by counting photons scattered within the source and thecamera. In contrast to 99mTc, in which the extra counts are due only to self-scattered photopeak photons, 177Lu high energy gammas also contribute to thecounts detected within the photopeak window. when TEW scatter correction wasapplied to planar acquisitions, the results for both isotopes were similar to thoseobtained from the tomographic data.The camera CF translates counts into activity, and therefore is different forevery isotope. 99mTc emits 0.89 140 keV photons per decay, while 177Lu emitsonly 0.1 of 208 keV photons per decay. Therefore, for the same activity of thesetwo isotopes, nine times more counts are expected for 99mTc than for 177Lu. sothe value of the 99mTc calibration factor should be greater by approximately thesame amount. Indeed, the ratio of the measured values is approximately 8.6. Thissmall difference can be attributed to the change of detection efficiency of the NaIdetector on photon energy. This indicates that corrections are being done in theappropriate way.In summary, our results suggest that planar scans of point sources placed inair are sufficient for the determination of the camera calibration factor. However,for planar scans, scatter correction should be applied in the same manner as it isdone in tomographic studies. This is particularly important for the 177Lu case inwhich background from high energy scattered photons can be substantial.974.4. Discussion4.4.2 177Lu QuantificationThe results of experiments performed in air are presented in Figure 4.3. In thiscase, the ROIs are big enough to avoid any partial volume effect (PVE)’s becauseall the activity, including spill out, is being counted when very low threshold of0.1% is used. Therefore, a fairly uniform behavior of the quantification is obtainedfrom both scatter correction methods for both spheres and bottles.APDI is an analytical method that is computationally demanding, so in orderto run it faster, it down-samples the attenuation map from a 128× 128 matrix toa 64× 64 . This down-sampling may be the cause of its slight overestimation ofactivity as down-sampled attenuation map results in increased values of attenua-tion coefficient in some voxels. Please note that this effect does not occur whenobjects are scanned in water.For the bottles placed on the camera bed (C1, C2, C3 and C4), although theresults from APDI are still overestimating the activity, the accuracy is better thanTEW. In this case the volumes were larger and scatter conditions are non-uniform,which is better corrected for by APDI.Figure 4.3b summarizes the distribution of quantification accuracy for bothscatter methods. For TEW, the mean accuracy is within 1% of the true value.However, the distribution is skewed to lower values, meaning that for TEW anunderestimation of activity is more likely to occur. This mostly occurs for bottlesplaced on the camera bed as TEW performs better when the objects are furtherfrom the scattering medium. APDI overestimates all the activity values, but is984.4. Discussionmore accurate in a non-uniform attenuation environment.Adding a scatter medium (i.e. water) showed the following trends. Under-estimation of activity for small objects (S1 and S2) which is most likely due toPVE’s as the 1% threshold for the ROI does not account for the activity spill outfrom these two very small spheres. In general, TEW overestimates the activitywhile APDI provides more accurate values. Images reconstructed with APDI aresharper than TEW, with smaller scattered photons tails around the spheres. In thethorax phantom (non-uniform attenuation case), both scatter correction methodsoverestimates the activity of the inserts (T1, T2, T3 and T4), but again, APDI ismore accurate. In particular, bottle T4 which was placed in contact with the beefbone is highly overestimated by TEW while APDI accounts for the true activityvery precisely. Finally, for the bottles placed between water bags (C1, C2, C3and C4), both scatter correction methods perform equally well as in this case, thebottles are relatively uniformly surrounded by water. The large error observed forbottle C1, both in air and water, indicates that most certainly there was an error indetermining its true activity.The box plot of Figure 4.4b shows that the accuracy distributions for bothTEW and APDI are skewed towards lower values. While TEW tends to overesti-mate the activity, APDI mostly underestimates it. This shift to lower values is dueto PVE’s in the smallest objects and the underestimation of activity in bottle C1.On average, APDI is more accurate than TEW and its variation about the meanvalue is smaller.The situation becomes more challenging when a hot background is present.994.4. DiscussionIn particular, sphere S1 has been removed from the analysis because of its smallvolume (1mL) it could not be distinguished from the background. For the samereason, the quantification results for sphere S2 (2mL), although reported here, arenot reliable.In the hot background case, the selection of the ROI cannot be performed bysetting a low threshold similar to the previous two cases, because that will includetoo much of background activity in the ROIs. The 40% threshold, which is oftenused in clinical studies [126], clearly is too high and does not properly account forthe spill out effect. For this segmentation method both TEW and APDI perform ina reasonably similar manner. Applying segmentation based on the true dimensionsof the objects (based on CT-images) shows strong PVE’s, most pronounced forsmall objects S2 and S3. The accuracy of the CT method for S4 was very similarto the 40% threshold, but for larger objects this method performs better than the40% threshold. However the CT based segmentation method is difficult to usein patient studies because boundaries of organs/tumors are not always visible. Ingeneral, the best results are obtained with the IADT segmentation method. Inparticular, for volumes larger than 34 mL, the accuracy of quantification remainsis better than 5% . This result indicates that in patient studies, in the determinationof kidney (critical organ) activity similarly good accuracy can be achieved.The box plot of Figure 4.5d compares the three segmentation methods andshows that the best accuracy is obtained with IADT. Because all experiments withhot background involved phantoms with uniform distribution of attenuation, thequantification accuracy of both scatter correction methods is very similar for all1004.5. Conclusionthree segmentation approaches. Nevertheless, all three segmentation methods un-derestimate the activity indicating the need for improvement in segmentation tech-niques.Our results suggest that the protocol described in this study is accurate in de-termining the activity of 177Lu, but also it shows that organ/tumor segmentationremains challenging.4.5 ConclusionSeveral and different phantom experiments were performed with inserts of differ-ent shapes, sizes, scatter, and attenuation conditions. We also performed Monte-Carlo simulations to understand the behavior of scatter in different energy win-dows and correctly account for it when using the APDI correction method. Fi-nally, we investigated the methods used for determination of the camera calibra-tion factor with planar and tomographic acquisitions. Our simulations indicatedthat contamination into the photopeak window from high energy photons, couldbe approximated by the self-scatter from the upper scatter window.When determining the camera calibration factor, our results showed that, aslong as TEW is applied to the planar scan, the same way as done in the SPECTreconstructions, the results were identical. This suggests that planar scans areenough for this purpose making it an advantage as they are easier to perform thantomographic acquisitions of phantoms.For phantoms scanned in air, quantification within 5% and 10% was achieved1014.5. Conclusionwhen performing TEW and APDI scatter correction methods, respectively.For phantoms scanned in water, APDI scatter correction provided more accu-rate determination of activity than TEW with errors within 5%. These experimentsvalidated our reconstruction algorithm showing that our quantitative reconstruc-tion is accurate. Also, our results suggest that APDI corrects better for scatter inchallenging situations when the objects are close to the camera bed or placed innon-uniform medium.When background activity was added, three segmentation methods were tested,40% fixed threshold, true volume of the objects, and our IADT method. In allcases the activity was underestimated. However, based on the scans in air andwater, the problem does not seem to be related to the reconstruction algorithmbut rather to the segmentation method. IADT provided the best quantificationaccuracy which was very similar for TEW and APDI correction. The differencesshown between the two scatter correction algorithms, seems to be compensated bythe fact that threshold curves were calibrated using set of corrections appropriatedfor each method. Although dosimetry for very small lesions is still a challengeand better segmentation algorithms are still required, IADT is a good methodwhen performing dosimetry for medium size organs like the kidneys. Finally, forlesions located in regions with non-uniform attenuation,the APDI scatter correc-tion method is recommended.102Chapter 5Deadtime Corrections5.1 IntroductionThe Committee on Medical Internal Radiation Dose (MIRD) [60] describes a pro-cedure in which the accurate quantification of activity and its temporal behaviorwithin the different organs of interest should be determined. The temporal behav-ior of the activity can be determined by scanning the patient several times, usuallyduring the first week after injection. Quantification is achieved when scatter andattenuation corrections are included in image reconstruction.In typical diagnostic scans, the activity injected to the patient is low resultingin low count rates so no DT corrections are needed. However, in therapy proce-dures (as those performed with 177Lu), patients are injected with activities in theorder of GBq resulting in high flux of gammas. In order to accurately quantify theactivity in these situations, DT corrections might be necessary.Several studies have investigated DT effects on Anger cameras with 99mTc and131I [128–133]. However, to the best of our knowledge, only two studies have ex-amined this effect for 177Lu radionuclide therapies dosimetry purposes. Beaure-gard et al. [107] studied a quantification protocol in which scatter correction using1035.2. Materials and Methodsthe TEW method was applied. Both the entire spectrum and the photopeak DTwere measured, and data points were fitted to the Sorenson’s paralyzable model[129]. They found a significant difference in the DT values for the entire spec-trum compared to the photopeak but applied DT corrections using counts fromthe entire energy spectrum. Celler et al. [134] proposed a new method for DTcorrection based on data acquired during patient scans. They suggested to place asmall marker in the FOV and image it simultaneously with the patient in order todetermine count losses with respect to the imaging of the small marker without thepatient. This procedure however, presents challenges in the clinical environment.The aim of this study was to investigate the DT effects that are present inimaging of 177Lu radionuclide therapies, and assess differences between the DTvalues determined using the entire spectrum and photopeak counts only.5.2 Materials and Methods5.2.1 SPECT Cameras, Collimators, and Energy WindowSetupThe experiments were performed using two SPECT cameras:1. A Siemens SymbiaT (Siemens Medical, Germany) with a MELP collimator.2. An Infinia Hawkeye (GE-Healthcare, UK) with a MEGP collimator.Two photopeak energy windows (PW) were specified for the 113 keV and 208 keVphotopeaks of 177Lu (Table 5.1). Additionally, lower scatter window (LSW), and1045.2. Materials and MethodsWindow Name Lower Limit [keV] Center [keV] Upper Limit [keV]LSW113 88.4 95.0 101.7PW113 101.7 113.0 124.3USW113 125.0 139.0 152.9LSW208 153.0 170.0 187.0PW208 187.2 208.0 228.8USW208 229.5 255.0 280.5Table 5.1: Energy window limits for the two photopeaks of 177Lu. Additionalenergy windows were used to collect data for the whole spectrum.upper scatter window (USW) were defined for each of the photopeaks in orderto perform TEW scatter correction. Furthermore the counts in the entire energyspectrum were also measured.5.2.2 Planar AcquisitionsActivity of 177Lu was diluted in water to obtain a solution with a concentrationof 35.7± 1.4MBq/mL. In order to measure the DT, the camera response has tobe determined over a wide range of activities. For this purpose, we filled twenty-five 10mL syringes with the prepared solution, and the activity contained in eachsyringe was measured using a CRC-25R (Capintec, USA) dose calibrator. Thesyringes were sequentially emptied into a 295 mL bottle located 5 cm off-centerinside a cylinder filled with water (Figure 5.1). The cylinder was placed on thecamera bed between the detectors, and the center of the bottle was positioned at35 cm and 25 cm from the collimators surface (Figure 5.1). Detector 2 was closerto the bottle and this allowed us to investigate effects of attenuation. Activityremaining in the empty syringe was measured again in the dose calibrator and the1055.2. Materials and MethodsFigure 5.1: Coronal view of the positioning of the phantom and the bottle insideit with respect to the two detectors.net activity added to the bottle was calculated for the 25 syringes.Measurements for the two cameras were performed on separate days. On day1 we performed measurements using the SymbiaT camera, while measurementsusing the Infinia camera were performed on day 2.On the first day, a total of 9.12± 0.91 GBq of 177Lu were distributed intothe 25 syringes. The first planar scan was performed for the empty bottle, andsubsequent scans were performed after emptying each syringe providing a total of26 scans.On the second day, the activity that was added to the bottle on day 1 wasredistributed back into the 25 syringes and the measurements were repeated.Energy spectra were also collected using the SymbiaT camera capable of ex-porting ASCII files containing the data. This was done for both detectors and alow (17.3 kcps), medium (101.5 kcps), and high (306.3 kcps) count rates. Thespectra were normalized such that the area under the curve was equal to 1, gen-erating probability density functions. The three different spectra were plotted onthe same axis to facilitate comparison.1065.2. Materials and Methods5.2.3 Data AnalysisTwo different methods to determine count rates were used:1. Count rate for the entire spectrum: The counts in the whole FOV of theimage were divided by the duration of each scan to determine the countrate.2. Count rate for the photopeak: TEW scatter correction [79] was applied toboth photopeaks. Scattered counts (Cs) were estimated from the counts inthe LSW (Cls)and USW (Cus) using equation 5.1. The widths of the LSW,PW, and USW are given by wls, wpw,and wus, respectively. The scattercounts were removed from the photopeak and the count rate was calculatedby dividing them by the duration of each scan.Cs =(Clswls+Cuswus)wpw2(5.1)The acquisition time varied from 3 minutes for very high activity to 5 minutes forlower activity points. In order to estimate the expected true count rate at highervalues of activity, a linear fit was made using to the data points corresponding toactivity in the phantom lower than 2.1GBq. We assumed that no deadtime waspresent at these count rates.Sorenson [129] claimed that gamma cameras behave as having a combinationof paralyzable and non-paralyzable components. Several studies from the sec-ond half of the last century [128, 135] found that, at least for the range of count1075.2. Materials and Methodsrates encountered in medical procedures, the camera behavior could be accuratelyapproximated by a paralyzable model. More recently, Silosky et al. [133] an-alyzed new cameras, including a Symbia and a GE Millennium, and concludedthat for count rates below 375 kcps, the cameras behaved as paralyzable systems.Guirardo et al. [132] found that for a Symbia camera, a sharp change in cameraresponse can be observed at very high count rates, but in the region below thischange both paralyzable and non-paralyzable models accurately fit the data.Based on our experience with patients treated by our collaborators at the Cen-tre Hospitalier Universitaire de Quebec, the count rates observed between one andfour hours after the injection correspond to the low range of the data discussedhere. Since we did not see the sharp change in behavior mentioned by Guirardo,we fitted our data to Sorenson’s paralyzable model [129].ro = rte−rtτp (5.2)where rt represents the true count rate, estimated from the linear fit, ro is theobserved count rate obtained from the experimental measurements, and τp is theparalyzable DT parameter. Equation 5.2 was linearized, the data was fitted using atotal square fit algorithm, and the slope of the line provided the desired parameterτp.deadtime correction factors (DTCF) were calculated as the percentage differ-ence between true and observed count rates:1085.2. Materials and MethodsDTCF =rt− rort×100 (5.3)The observed count rate value was obtained from the experiment while thevalue of the true count rate was calculated using equation 5.2. Note that thisequation cannot be solved analytically and the solutions are given by the LambertW function. Numerical values were obtained using Matlab (Mathworks, USA)and plots of DTCF vs. observed count rate were made for both cameras.Finally, as we collected data using different energy windows, we also com-pared the DT corresponding to the entire spectrum with that obtained for the twophotopeak windows.An approach to determine the DT for the PW if the DT for the entire spectrumis known has been discussed by several authors [128, 136]. The method involvescalculating the window fraction and relating it to the full spectrum DT τ f s.Window fraction w f is the fraction of counts detected within a specified win-dow, relative to the total number of counts in the entire energy spectrum. Everyphoton reaching the detector can cause DT even if it is not counted in the specifiedenergy window. Narrower energy windows, will see an increase in DT as the pro-portion of events detected outside of this PW will be higher, reducing the windowfraction [61, 129]. It has been proposed that the observed DT for the specifiedenergy window τw is related to the entire spectrum deadtime by:τw =τ f swηf(5.4)1095.3. Resultswhere η is a positive constant. Cherry et al. [61] mention a value of η = 1,while Silosky et al. [133] report values of η = 1.4.To validate if equation 5.4 can be applied in the situation in which scatter isremoved from the PW, we calculated the window fraction of the two PW’s forboth detectors. We used all the data points measured and calculated an averagewindow fraction with a standard deviation. We then used equation 5.4 with thevalues of η = 1 and η = 1.4 from the literature, and compared the resulting τwwith those which were obtained in our experiments.5.3 ResultsFigure 5.2, shows the spectra collected for both detectors for the Siemens Sym-biaT camera at low, medium, and high count rates. The vertical dashed lines showthe positions of energy windows. In both photopeak windows, the lowest countrate shows the highest photon intensity. In the scatter windows, the behavior isopposite, with the highest count rate showing the highest intensity of scatteredphotons.Comparing the detectors, it can be seen that the intensity of the 113keV peakis approximately 1.3 times higher in detector 2 with respect to detector 1, whilethe 208 keV peak has an intensity of 1.1 times higher in detector 2. The proportionof Pb X-rays (approximately 70 keV) from the collimator with respect to charac-teristic X-rays of 177Lu (approximately 50 keV) is greater for detector 1 whichwas closer to the bottle filled with activity.1105.3.ResultsEnergy [keV]0 50 100 150 200 250 300 350 400Probability Density0102030405060708090-310×Measured Spectra for Different Count Rates17.3 kcps101.5 kcps306.3 kcps113LSW113PW113USW208LSW208PW208USW(a) Detector 1Energy [keV]0 50 100 150 200 250 300 350 400Probability Density0102030405060708090-310×Measured Spectra for Different Count Rates17.3 kcps101.5 kcps306.3 kcps113LSW113PW113USW208LSW208PW208USW(b) Detector 2Figure 5.2: Measured spectra recorded at different count rates for both detectors.1115.3. ResultsThe observed count rates vs. the true count rates for the entire spectrum andthe two photopeaks of 177Lu are presented in Figure 5.3. Results for both theInfinia and SymbiaT camera are shown. Each detector has been plotted separatelywith corresponding τp. The dashed lines represent the identity. All the symbolsin blue show the data for detector 1, while those in black are for detector 2. Thefit to the paralyzable model is displayed in red. In the case of detector 2 of theSiemens camera, the values of DT determined from PW208 and PW113 (with TEWscatter correction applied) exceeded the DT values determined from the entirespectrum by a factor of 25 and 32, respectively. The results were similar for theInfinia camera in which the values were 11 and 6 times higher for the PW208 andPW113, respectively, relative to the full spectrum. Table 5.2 shows the coefficientof determination (R2) that suggest a very good fit.SymbiaT InfiniaDet 1 Det 2 Det 1 Det 2Full Spectrum 0.996 0.980 0.992 0.998113 keV Photopeak 0.835 0.999 0.990 0.943208 keV Photopeak 0.968 0.999 0.983 0.923Table 5.2: Coefficient of determination (R2) of the fits of the data to the paralyz-able model.Figure 5.4 shows the DTCF obtained for both cameras as a function of theobserved count rate. Both cameras show higher DTCF values for the photopeakwindows compared to the full spectrum case. For the Symbia camera, the highestDTCF was observed in the PW113 with up to 29%, about 3.6 times greater than thatfor the full spectrum in the measured range. The PW208 reached a maximum of1125.3. ResultsTrue Count Rate [cps] ×1050 0.5 1 1.5 2 2.5 3 3.5 4Observed Count Rate [cps]×10500.511.522.533.54D2: τp= 0.19 ± 0.18 µsD1: τp= 0.40 ± 0.25 µsFull SpectrumDetector1Detector2     Fit(a) Siemens Full SpectrumTrue Count Rate [cps] ×1040 5 10 15Observed Count Rate [cps]×104051015D2: τp= 2.27 ± 0.11 µsD1: τp= 3.14 ± 0.15 µsFull SpectrumDetector1Detector2     Fit(b) GE Full SpectrumTrue Count Rate [cps] ×1040 0.5 1 1.5 2 2.5 3 3.5 4Observed Count Rate [cps]×10400.511.522.533.54D2: τp= 4.79 ± 0.18 µsD1: τp= 4.43 ± 10 µsTEW PW208Detector1Detector2     Fit(c) Siemens PW208True Count Rate [cps] ×1040 0.5 1 1.5 2Observed Count Rate [cps]×10400.20.40.60.811.21.41.61.82D2: τp= 23.88 ± 0.94 µsτ= 56.10 ± 1.5 µsTEW PW208Detector1Detector2     Fit(d) GE PW208True Count Rate [cps] ×1040 0.5 1 1.5 2 2.5 3 3.5 4Observed Count Rate [cps]×10400.511.522.533.54D2: τp= 6.04 ± 0.23 µsD1: τp= 10.16 ± 38 µsTEW PW113Detector1Detector2     Fit(e) Siemens PW113True Count Rate [cps] ×1040 0.5 1 1.5 2Count Rate [cps]×10400.20.40.60.811.21.41.61.82D2: τp= 14.43 ± 0.63 µsD1: τp= 27.19 ± 0.9 µsTEW PW113Detector1Detector2     Fit(f) GE PW113Figure 5.3: Observed count rates as a function of true count rates. The values ofτ from the paralyzable model are shown for both detectors and both cameras.1135.3. Results0 1 2 3 4051015202530DTCF [%]Observed Count Rate [cps] Full Spectrum TEW PW208TEW PW113x104Siemens Symbia DTCF(a)0.0 0.5 1.0 1.5 2.0010203040506070GE Hawkeye  DTCFObserved Count Rate [cps]x104DTCF [%] Full Spectrum TEW PW208 TEW PW113(b)Figure 5.4: Deadtime correction factors for observed count rates in a SiemensSymbia and GE Hawkeye SPECT/CT cameras.22% DTCF. For the Hawkeye camera the highest DTCF occurred for the PW208with a change in behavior at around 1.5× 104cps for which the DTCF reached57%. At this count rate, the DTCF were of 26% and 4% for the PW113 and thefull spectrum, respectively. At the count rate of 2× 104cps, the Symbia camerashowed a full spectrum DTCF of 3.9%, basically the same as Hawkeye. However,for the PW’s the DTCF values were higher for the Hawkeye camera even thoughthe count rate was lower due to the experiment being performed one day later.Table 5.3 summarizes the window fraction for each photopeak, detector, andfor each manufacturer’s camera. For both photopeaks, detector 2, which wascloser to the bottle containing activity, showed higher window fraction than de-tector 1. GE’s Infinia camera also showed a higher window fraction for bothphotopeaks compared to Siemens’ Symbia.The values presented on the first four columns of Table 5.4 show the DT valuesfor each photopeak window calculated using η found in the literature [61, 133].1145.3. ResultsSiemens GEDet1 Det2 Det1 Det2wf113 [%] 16.41±0.18 19.24±0.26 19.95±0.28 22.34±0.42wf208 [%] 12.18±0.24 13.89±0.15 14.52±0.54 15.95±0.47Table 5.3: Window fractions (in percentage) for the two photopeaks, detectors,and SPECT cameras.Siemens GE η to achieve fitted value of τwη = 1 η = 1.4 η = 1 η = 1.4 Siemens GEτw113 [µs] 1.0 1.9 10.2 18.5 2.1 1.2τw208 [µs] 1.4 3.0 14.2 29.7 1.6 1.3Table 5.4: Values for τw obtained with equation 5.4 using values of η available inliterature. Window fractions obtained from detector 2 were used as it covered alarger range of count rates. The last two columns show the value of η required tocorrectly obtain the DT value of our experiments.Only detector 2 window fractions were used, because this detector covered a widerrange of count rates and the fit to the paralyzable model showed higher coeffi-cients of determination than that for detector 1. The last two columns of the tablepresent the value of η that is required to correctly obtain the value of DT of ourexperiments.The values of the DT for the Siemens camera were underestimated when usingη from the literature, and higher values of this parameter were needed to obtainthe measured τw.For GE, η = 1 underestimated τw and η = 1.4 overestimated it.Although this is not a rigorous test for the validity of equation 5.4, it shows thatcare has to be taken when applying this equation as it does not correctly predictDT when TEW scatter correction has been used.1155.4. Discussion5.4 Discussion5.4.1 Energy Spectrum at Different Count RatesThe spectra collected at different count rates show various interesting effects thataffect DT. The reduction of intensity of both photopeaks (i.e. PW113 and PW208)as the count rate increases can be explained by1. count losses caused by processing time by the camera electronics2. pileup effect. Zasadny et al. [136] found the pileup occurring when per-forming a DT study with 99mTc and 131I. Two or more photons that reachthe detector at the same time may be counted as a single photon with the en-ergy equal to the sum of the energies deposited by the photons. This effectcauses an increase in probability of detecting photons in the USW’s (Fig.5.2). Detector 2 showed this effect at a higher degree than detector 1.The effect of attenuation was also observed in the lower energy ranges of the spec-tra where production of X-rays (from Pb and Lu) are highly attenuated. Again,this affects the spectrum of detector 1 to a higher extent as the characteristics X-rays of 177Lu need to travel a longer distance in water before reaching the detector.As the energy of photons increases, the linear attenuation of water decreases. Thisresults in a lower chance of photon-interaction with the water for high energy pho-tons. The intensity of the higher photopeak (208 keV) in both detectors has almostthe same intensity, while the 113 keV in detector 1 is lower than in detector 2.1165.4. DiscussionThe scatter windows showed slightly higher intensities for the highest countrate which is again explained by pileup effects.5.4.2 Deadtime ValuesSiemens SymbiaTThe first thing to note from the Siemens camera data presented in Figure 5.3, isthat the behavior of both detectors is very similar. When the curves were fittedto the paralyzable model, detector 1 always showed a higher uncertainty in τp,which is most likely due to the fact that, due to attenuation, the maximum countrate measured by this detector corresponded to smaller τ values. Although thevalues of DT measured for the entire spectrum and for the photopeak windowscorrected for scatter are equal within experimental uncertainties, we decided touse values from detector number 2 in the rest of our analysis.The differences in DT between PW’s and full spectrum, although high, are notsurprising. Silosky et al. [133] also found a factor of 20 difference for 99mTc be-tween the DT determined based on the full spectrum and the photopeaks. Themain cause for these differences is the pileup effect and scatter. Although thewhole spectra display count losses due to the camera electronics, the PW’s aremore affected by pileup effects. When photons are detected out of the specifiedrange, this gives the impression of missing photons and thus generating a higherDT.For the SymbiaT camera, the effect is higher for the PW113 (Figure 5.4), most1175.4. Discussionlikely due to increased scatter of low energy photons and higher efficiency ofthe detector at low energies. Both these effects can increase the pileup. Thisis also a reason why camera DT is radioisotope dependent. Different peaks fordifferent isotopes have different energies and thus different detector responses.As mentioned before, previous studies suggest that better quantification of 177Luactivity is achieved when data from the PW208 is used. The fact that less deadtimeoccurs for this 208 keV peak is another reason for which it should be the one usedfor quantification of 177Lu.Although the paralyzable model showed good fit to the data when trying touse equation 5.4 to predict the values of τw from the window fractions, the valuesfound in literature for the exponent η underestimated the DT values. The reasonsfor this are not completely clear, η = 1 and η = 1.4 were obtained for photopeakwindows that were not corrected for scatter. When applying TEW scatter cor-rection, different values of η were needed to correctly predict DT for the scattercorrected photopeak. However, we are not establishing a model to calculate theDT for scatter corrected photopeaks as this would require further investigationsand new experiments. On the other hand, if the protocol used to quantify 177Luactivity involves the same energy windows and scatter correction is applied thenthe DT values and DTCF that we calculated should provide accurate correctionfor missing counts.1185.4. DiscussionGE HawkeyeThe measurements of count rates for the Hawkeye camera were performed oneday after Symbia. After filling the syringes on the second day, and performingthe first scan with the empty bottle, we found out that approximately 1.9 GBqhad precipitated into the walls of the bottle and could not be extracted. Due tothis effect, we could not measure some of the low range data points. Detector1 detected lower count rates due to the higher attenuation of photons from thesource to it. One possible solution to the problem was to fit all the data points fromboth detectors with only one curve. This however, would have not let us observethe different conditions of attenuation on deadtime that we obtain by placing thebottle containing activity off-center withing the phantom. However, contrary tothe Siemens case, this camera showed a much different behavior between the twodetectors. This was most likely caused by the difference in the behavior of thephotomultiplier tubes and the electronics within each detector. This did not allowus to perform one fit for all the points.Big differences were also observed in the DT values determined based onthe PW’s compared to the full spectrum. In this case however, the highest lossof counts was seen in the PW208. This effect, however, seems to be caused byproblems with fitting the paralyzable model without data points at low count rates.The red lines in Figure 5.3, which represent this fit, underestimate the observedcount rate for detector 2 which is the one used for the calculation of the DTCF. TheDTCF plot (fig. 5.4) shows a change in behavior at 1.5×104 cps. This suggeststhat the paralyzable model is not correctly describing the data, and that either the1195.5. Conclusioncamera behaves as non-paralyzable system, corresponds to a combination of bothparalyzable and non-paralyzable components, or that the missing points in thegraph did not allow us to correctly describe the behavior of the Hawkeye camera.5.5 ConclusionDeadtime measurements for a Siemens SymbiaT and GE Hawkeye SPECT cam-eras have been performed by adding activity into a bottle placed off-center in acylindrical phantom filled with water. Plots of observed count rate vs. true countrates were fitted to a paralyzable model and deadtime values (τ) were determinedfor each detector and for the full spectrum and the two photopeak windows setaround the 113 keV and 208 keV gamma emissions of 177Lu. The paralyzablemodel showed to be appropriate for the range of counts studied for the Siemenscamera. The count rate in the photopeak windows was calculated by removingscatter using the triple energy window scatter correction method. Prediction ofphotopeak window deadtime based on the window fraction did not follow mod-els found in literature most likely because those models do not account for thesubtraction of scatter presented here. We then develop deadtime correction plotsbased on observed count rate and higher deadtime was observed for the photopeakwindows. This suggests that deadtime corrections should be performed based onscatter corrected photopeak window and not by using the deadtime determinedfrom the full spectrum. Additionally, the deadtime values are lower for the 208keV photopeak, therefore it is recommended that 177Lu quantification should be1205.5. Conclusionbased on acquisition of the 208 keV photons. The deadtime correction should beapplied to the reconstructed image before converting counts to activity values.121Chapter 6Dosimetry in Patients UndergoingPRRT for NETs Treatment6.1 IntroductionThe purpose of every radiation treatment is to kill all tumor cells while avoiding tokill normal tissue cells. The radiation therapy oncology group (RTOG) have beentrying to standardize toxicity to avoid any possible normal tissue complicationswhen performing conventional external beam radiation therapy [137]. Guidelinesfrom RTOG, are based on previous clinical experience, and are specifically estab-lished for each tumor type [138].In general, cell response depends on: [137]• Repair of DNA damage.• Re-assortment (i.e. cell cycle).• Re-population.• Re-oxygenation as hypoxic cells are less sensitive to radiation.1226.1. IntroductionAdditionally, Pouget et al. [139] discusses how biological effects depend on fac-tors that include:• Dose rate• Dose fractionation• Volume of irradiated tissueWhile the linear-quadratic model [140] has been a widely recognized standardto assess biologic responses to radiation dose in external beam radiotherapies[138, 139, 141], it may not necessarily reflect those in PRRT [139]. The ab-sorbed dose rate and the spatial energy deposition in PRRTare different than thosefor external beam therapies. This could be related to the fact that external beamtherapies are done with a high absorbed dose rate, while PRRT have low absorbeddose rates. Irradiation with external beams is also more homogeneous than inPRRT. Additionally, internal therapies with alpha particles or Auger electronshave a higher linear energy transfer than X-rays or electrons used in externalbeam. However, for the case of 177Lu, the linear energy transfer should not bevery different when compared to X-rays used in external beam therapy, but theabsorbed dose rate is lower. [139]The big problem when trying to deal with radio-biology and the applicationof survival models in PRRT is the limited availability of dosimetry data in clinicalstudies. Data from these studies are usually presented in values of injected activity,injected activity per patient’s mass, or injected activity per patient’s body surface.Due to this deficiency, it has been difficult to determine the validity of the linear1236.1. Introductionquadratic model in PRRT. The absorbed dose and dose rates are parameters forwhose effect on response to radiation still need to be understood in PRRT. To beable to identify these relationships, an accurate internal dose assessment procedureneeds to be developed. [139]The previous chapters have focused on the accurate determination of the ac-tivity distribution in order to determine the cumulative activity which is necessaryfor dose calculation (equation 2.20). There are several approaches to calculatepatient dose:• Organ Level: In this approach, it is assumed that the activity is uniformlydistributed within the source organ and the dose is deposited uniformly inthe target organ.• Sub-Organ Level: Different regions of the organs can have a different be-havior for the uptake and washout of activity. Some regions can have higheruptake while others can hold the activity for a longer time. Multi regionmodels [142] can then be used to perform dosimetry for each of the differ-ent regions instead of assuming the same dose distribution in the organ.• Voxelized: In this case the activity distribution provided by the SPECT/CTquantification, which involves different values for each voxel, is used tocalculate doses for each individual voxel.As mentioned previously in Chapter 2, in order to determine the cumulative activ-ity, the activity distribution at different times after injection should be measured.However at this point, little has been said about the S-factor in equation 2.20 which1246.1. Introductioninvolves the physical characteristics of the radioisotope, that combined with thebiokinetics, determine the dose to the target.6.1.1 S-values at the Organ LevelThe most commonly used dosimetry approach is the organ level internal doseassessment (OLINDA) [143]. In this approach, S-values are pre-calculated usingMC in standard phantoms that represent a standard male or female. The shapesand sizes of organs and the patient geometry are defined in these pre-establishedphantoms thus does not provide a personalized dose assessment. The activity isassumed to be uniformly distributed in the source organs, and the absorbed dosewithing the target organ is also uniformly distributed.A newer version of OLINDAcontains S-values pre-calculated for more accurate phantom geometries that betterapproximate the human body [144].6.1.2 S-values at the Voxel LevelIn the voxelized approach [145, 146], kernels of deposited doses are generatedfrom single voxels of unit activity. These kernels are pre-calculated using MC andare unique for each radioisotope, voxel size, and tissue type. A limitation of thismethod is that a single tissue type is modeled and tissue inhomogeneities are notconsidered.The S-value kernels must then be convolved with the cumulative activity dis-tribution in order to obtain the absorbed dose distribution.1256.1. Introduction6.1.3 Full Monte-Carlo ApproachIn a full MC approach, a reconstructed image containing the activity distributionof the patient is used as the input for the MC code, and CT or µ-map is used to tomodel energy deposition by radiation within the patient. The result is a 3D dosedistribution, which is patient specific as it is calculated using the geometry andactivity distribution of each patient.6.1.4 JADA Dosimetry PackageGrimes et al. [59] developed a graphical user interface (GUI) to calculate OLINDAand voxelized dose distributions. As part of the software, voxel S-value kernelshave been pre-calculated for 177Lu, with voxel size corresponding to those in thereconstructed images of our data. These kernels were used for our voxel leveldosimetry.6.1.5 AimThe aim of this chapter is to compare doses calculated using the organ level andvoxelized dosimetry approaches on patients undergoing PRRT for the treatmentof NETs. The quantification of activity of 177Lu was performed using methodspresented in the previous chapters.1266.2. Methods6.2 MethodsFour patients undergoing PRRT were scanned three or four times in a week fol-lowing the 177Lu-DOTA-octreotate injection. The mean injected activity was7.1± 0.8 GBq. Figure 6.1 shows images of these patients on the day of injec-tion, one day after, and three days after injection.The SPECT scans were performed at the L’Hotel-Dieu de Quebec site of theCHU de Quebec – Universite Laval center (Quebec City, Canada), using a Sym-biaT (Siemens, Germany) SPECT/CT camera with the energy window setup pre-sented in Table 4.1, and a total of 90 projections. Images were reconstructedusing the OSEM algorithm using attenuation correction (AC), and triple energywindow (TEW) scatter correction method. Deadtime corrections were applied asdescribed in chapter 5 using τ = 4.79µs which was the value obtained for theSiemens’ camera for 208 keV photopeak when TEW was applied (Figure 5.3).Plots of activity as a function of time (time-activity curves TAC) were createdfor the left and right kidneys of the patients using the activities determined fromthe reconstructed images at the different time intervals. The curves were fittedto a mono-exponential model in order to determine the effective half life of theradiopharmaceutical. As most of the patients were only scanned three times, fitsto bi-exponential models were impossible to perform.Three methods were used to determine the dose to the kidneys:1. OLINDA software: In this model, the residence time (i.e. cumulative ac-tivity divided by injected activity) calculated from the time-activity curves1276.2. Methodswas used as an input to OLINDA. It was assumed that the dose to the kidneywas from self-contribution so only data for this organ was used. The doseestimate was performed using the male phantom from OLINDA as all the 4scanned patients were males. None of the default values in OLINDA weremodified. The output provides the dose per injected activity.2. S-value kernel for organ level dosimetry: In this case an S-value kernelfor 177Lu available from JADA [59] was used. The cumulative activitywas calculated by determining the biokinetics in which the total activity inthe kidney was assumed to be uniformly distributed throughout its volume.The kernel was convolved with the image of the cumulative activity of thekidney, and a distribution in which each voxel had the same absorbed dosewas obtained.3. S-value kernel for voxelized level dosimetry: The same S-value kernelas used in method 2 was used here. The biokinetics from method 2 wasassumed to be the same for each of the voxels of the activity distributionobtained on the first scan after injection. In this case, the activity was notassumed to be uniformly distributed in the volume. After convolving thekernel with the cumulative activity of each voxel, a 3D dose distributionwas obtained.Segmentation was performed using the IADT because the results from Chapter 4showed that this method provided the best quantification of activity.1286.2. Methods(a) Patient 1- t0 (b) Patient1- t1 (c) Patient1-t3(d) Patient2-t0 (e) Patient2-t1 (f) Patient2-t3(g) Patient3-t0 (h) Patient3-t1 (i) Patient3-t3(j) Patient4-t0 (k) Patient4-t1 (l) Patient4-t3Figure 6.1: Images of the four patients scanned on injection day t0, one day afterinjection t1, and three days after injection t3.1296.3. Results6.3 ResultsFigures 6.2 to 6.5 show the time activity curve (TAC)’s and the dose-volume his-togram (DVH)’s for the organ level and voxelized level dosimetry. Patient 3 hadonly one kidney. The TAC’s show an effective half-life, based on the mono-exponential fit, for each kidney. The coefficient of determination is also shownfor each of the plots.The DVH’s have been plotted for dosimetry performed using methods 2 and3. The y-axis represents the percentage of volume that received a minimum of agiven dose. As the activity distribution was assumed to be uniform for method2, the minimum dose for 100% of the volume is lower than in the case in whichmethod 3 was used to determine dose. For this reason, the scale on the dose axisis different for each method.1306.3. Results(a) (b)(c) (d)(e) (f)Figure 6.2: Patient 1 TAC and DVH.1316.3. Results(a) (b)(c) (d)(e) (f)Figure 6.3: Patient 2 TAC and DVH.1326.3. Results(a)(b)(c)Figure 6.4: Patient 3 TAC and DVH.1336.3. Results(a) (b)(c) (d)(e) (f)Figure 6.5: Patient 4 TAC and DVH.1346.4. DiscussionOLINDAVoxelizedOrgan LevelIndividualVoxelsPatientLK[Gy]RK[Gy]LK[Gy]RK[Gy]LK[Gy]RK[Gy]1 2.80 1.61 2.93 2.97 3.02 3.132 1.02 1.37 1.19 1.60 1.44 1.803 3.63 - 4.66 - 4.79 -4 3.93 3.52 4.87 4.89 4.85 4.69Table 6.1: Dose to the kidneys for four patients undergoing PRRT estimated usingthree different methods.OLINDAVoxelizedOrgan LevelIndividualVoxelsPatientInjectedActivity[GBq]LK[ GyGBq ]RK[ GyGBq ]LK[ GyGBq ]RK[ GyGBq ]LK[ GyGBq ]RK[ GyGBq ]1 7.28 0.39 0.22 0.40 0.41 0.41 0.432 7.39 0.14 0.19 0.16 0.2 0.20 0.243 5.92 0.61 - 0.79 - 0.81 -4 7.75 0.51 0.45 0.63 0.63 0.63 0.61Table 6.2: Normalized dose per injected activity for the kidneys of four patientsundergoing PRRT.Table 6.1 shows the mean doses delivered to the left kidney (LK) and rightkidney (RK) calculated using the three approaches discussed in section 6.2. Table6.2 shows the doses delivered to the kidneys, normalized by injected activity.6.4 DiscussionThe results presented in the last section show that TAC’s do not necessarily followa mono-exponential behavior. While for patient’s 4 right kidney the coefficient of1356.4. Discussiondetermination R2 of the fit is very close to unity, in other cases the fit is not thataccurate. As expected, the effective half-life Te f f for all patients is lower thanthe 6.6 days physical half-life of 177Lu, indicating that the radiopharmaceuticalis removed from the kidney faster than just by the radioisotope physical decay.All the kidneys, inter-patient or intra-patient, showed differences in the effectivehalf-life of the isotope.Regarding the DVH’s, the differences between the shapes of the plots betweenthe organ-level approach (middle row of Figures 6.2 to 6.5) and the single-voxelapproach (bottom row of Figures 6.2 to 6.5) is due to the method in which activityis assumed to be distributed in the organ. In the former case, each voxel withinthe kidney is assumed to have the same activity so the dose is also uniformlydistributed around the organ and this is shown by the horizontal line at 100% ofthe volume with the sharp drop at the dose delivered. However, in the single-voxel case, where activities are considered at the voxel level, the histogram is ahorizontal line up to a certain value of dose and then slowly decreases showingthat the dose is not uniform within the organ . Voxels of the boundary of the organreceive less dose as radiation comes from only one direction, while voxels locatedin the center of the kidney receive radiation from all directions and absorb a higherdose. Due to this effect, the DVH shows that there are voxels that received up to3 times higher dose than that observed in the organ-level histogram.The average doses presented in Tables 6.1 and the uptakes per injected ac-tivity shown in Table 6.2 are lower for OLINDA than doses calculated using theother two methods. As default values of OLINDA were not modified (e.g. patient1366.5. Conclusionsweight or organ masses), these differences can be due to errors in estimating themasses of organs. The mass of the organ in the voxelized cases, was calculatedby knowing the dimensions of a voxel, taking the volume of the ROI obtainedfrom segmentation, and using the density of water. The voxelized organ-leveland individual voxels approach also showed differences, but these were not asbig when compared to OLINDA. In all the cases except patient’s 4 right kidney,the individualized voxels dose was higher and this is most likely due to the in-homogeneous distribution of dose where some regions of the organ have higheractivity. These results confirm the difference in uptake between patients and be-tween left and right kidneys, as shown by the TAC’s. While patient 3, which onlyhad one kidney showed an uptake of 0.81 Gy/GBq for the individual voxels case,patient 2 only had 0.2 Gy/GBq. Lastly, assuming that the maximum dose that canbe delivered to the kidney to prevent toxicity is equal to 23Gy, or 5.75 Gy pertreatment session, the results presented here suggest that all these patients wereunder-treated as none of them achieved this upper limit.6.5 ConclusionsThe analysis of dose calculations presented in this chapter show that personalizeddosimetry should become routine as not only the behavior of the activity in thekidneys was different for each patient, but the limits set for toxicity were notreached.137Chapter 7Single Photon Emission QuantitativeTomographic Reconstruction(SPEQToR)7.1 IntroductionThis chapter presents a graphical user interface (GUI) developed in Matlab (Math-works, USA) that includes several algorithms necessary to reconstruct SPECTdata with corrections required to create quantitative images. The GUI has beengiven the name of single photon emission quantitative tomographic reconstruc-tion (SPEQToR).SPEQToR was created using a series of routines programed in C++, and thegraphical interface environment available in MATLAB. It runs on a Linux envi-ronment capable of running C shells. The GUI is divided into several sections:1. Loading of the data: Options are provided to load DICOM files of nuclearmedicine data (projections),µ-map, and CT data. The program identifies themodel of the camera used for acquisition by reading the information from1387.2. Loading the Datathe DICOM file header and selects it automatically. It has been tested witha Philips Brightview, Siemens InfiniaT, and GE Hawkeye cameras. Addingmore camera models is possible and easy to implement.2. The section allows the user to display the data loaded in the previous step.3. The third section lets the user choose the reconstruction protocol or allowsto create a new one and save it for future use. The specification of a protocolincludes the selection of the window desired for reconstruction, the type ofcollimator used during acquisition, the reconstruction algorithm, options touse resolution recovery, type of scatter correction, attenuation correction,background subtraction, and the number of subsets and iterations for theOSEM reconstruction algorithm.4. The last section displays the reconstructed images and provides a simpletool to calculate the camera normalization factor.7.2 Loading the DataData obtained from modern SPECT/CT cameras use the DICOM standard[147]as a file format. Although supposed to be a standard, different manufacturers havemade changes to their DICOM files and some information contained in the headerhas either made private or is stored with different names. These different optionshave been implemented in the GUI code for the camera models tested. Figure7.1 shows an example of the functions which allow the user to load the data .1397.2. Loading the DataFigure 7.1: “Load Data” section of the GUI. This image shows a Nuclearmedicine and µ-map loaded from a Siemens camera. (The camera model is shownas specified in the DICOM file header.If the camera is not in the list, a new model should be added by selecting the“Not_Listed” option in the popup menu.Acquisition is usually performed with several energy windows that are usedfor scatter correction. However, the projection data corresponding to each energywindow can either be contained in a single or in separate files. When selecting thedata to load, the program asks for this information. When loading the µ-map, theuser selects the corresponding DICOM file. If the user decides to load CT images,a checkbox to specify if all the slices are included in one file (volume file) or ifseveral files are present (one file for each slice) has to be specified. An option todisplay the header of the nuclear medicine file is available and the user can checkacquisition parameters (see Figure 7.2).1407.2. Loading the DataFigure 7.2: Example header of a nuclear medicine data file.1417.3. Data DisplayFigure 7.3: Nuclear Medicine projections displayed on the left, attenuation mapshown in the middle, and the optional CT slices at the right.7.3 Data DisplayFigure 7.3 shows the “display” section of the GUI. The image on the left displaysthe projections data, with the total number of acquired projections shown at thetop. A slider at the bottom allows to scroll between projections, and the projectionnumber is shown in the box on top of the axis. Additionally, a second slider allowsto scroll through the different energy windows and the minimum and maximumenergy limits of the displayed window are shown below the sliders. A buttonbeside it, “QC Tests” opens a separate GUI in which quality control tests canbe performed. This GUI is shown in Fig. 7.4 and it displays the sinograms andlinogram of the collected data. With the sliders at the bottom, the user can easilyscroll through the different slices to detect possible problems in data collection.The image in the middle shows the attenuation map for the slice number dis-played in a text box at the top, and a slider at the bottom can be used to scroll1427.3. Data DisplayFigure 7.4: Quality control GUI to investigate the linogram and sinogram of thedata.through different slices. In some camera models, it has been observed that the µ-map is flipped with respect to the projection data. As the positioning of the map isnot being done with the information from the DICOM header, it is assumed to beregistered with the projection data. However, cases in which the first slice of theattenuation map corresponds to the last slice of the coronal view of the projectionsmake it necessary to flip the attenuation map. To correct for this, a button to flipthe map has been added. Options for displaying axial, coronal, and sagittal viewscan be selected for the µ-map and CT images. The image at the right shows theCT data.1437.4. Specifying the Reconstruction ProtocolManufacturer Collimator TypeSIEMENS LEHS, LEAP, LEHR, LEUHR, LEFB, MELP, HE, UHEGE LEHR, MELPADAC LEHRPHILIPS LEHRTable 7.1: List of collimator specifications currently available in the GUI. Morecollimators can be easily added by editing a file text and inputting the manufac-turer, type of collimator, hole diameter, length, and septal thickness of the colli-mator.7.4 Specifying the Reconstruction ProtocolThe reconstruction protocol is specified by selecting the algorithm for the recon-struction with or without resolution recovery (with the parameters of the collima-tor used during data collection), scatter and attenuation corrections, parametersof the OSEM algorithm, and in case of the analytical scatter correction, the pre-calculated lookup tables.The first step is to confirm the selection of the collimator, which is importantfor the resolution recovery correction. Table 7.1 shows the collimators that havealready been incorporated into the GUI and for which the dimensions have beentaken from the data-sheets of the manufacturers. For RR, the camera responsehas been modeled by a 2D depth dependent Gaussian function. The details of theprocedure are given in [148].Two reconstruction algorithms have been implemented: the filtered back pro-jection (FBP) and OSEM[74] in both 2D and 3D. When OSEM in 2D is used,each slice is reconstructed separately and no RR is applied, while for the caseof OSEM in 3D, the volume is reconstructed. RR is applied to the latter case if1447.4. Specifying the Reconstruction ProtocolFigure 7.5: Section of SPEQToR to define the reconstruction parameters.1457.4. Specifying the Reconstruction Protocolrequested by the user.Attenuation correction is selected via a checkbox under the popup menu toselect the scatter correction method (Fig. 7.5). The µ-map used for the attenuationcorrection must be loaded prior to the reconstruction.Three scatter correction methods have been implemented. They include broadbeam (BB) or modified attenuation correction[149, 150], TEW[79], and our APDI[80,81] method. The BB method uses lower values for the linear attenuation coeffi-cient, and scatter is compensated by under-correcting for attenuation. TEW re-quires the user to acquire the data with 2 (lower scatter and photopeak) or 3 (lowerscatter, photopeak, and higher energy scatter) windows. In case only 2 windowsare present, the higher energy scatter is assumed to be zero. The GUI sorts thewindows from lower energy to higher energies in a successive way, thus whenTEW is applied, the windows before and after the selected photopeak are used forthe correction.APDI analytically calculates scattered photons based on the energy windowof the photopeak. Pre-calculated lookup tables are required. A “generate lookuptables button” (Fig: 7.5) allows to generate these tables for a specified isotopeand energy window (if they have not been already pre-calculated). Otherwise, theuser should select the lookup tables indicated for the current isotope and energywindow peak.The attenuation correction is incorporated into the system matrix of the OSEMequation, while scatter (S) and background (B) projections are included in the de-nominator as explained in detail by Shcherbinin et.al. in [151] (equation 2.19).1467.5. Running the Reconstruction and Determining the Normalization Factor of the CameraNote that the contamination from high energy photons should only be used whenthe APDI scatter correction method is used as TEW already accounts for contam-ination from higher energy photons.Once the protocol is specified it can be saved for future use. Then, whenselected, a summary of its parameters is presented at the bottom of the availableprotocols list. An example of this is shown in Figure 7.6.7.5 Running the Reconstruction and Determiningthe Normalization Factor of the CameraAfter specifying the reconstruction protocol, the user should press the green but-ton “Reconstruct” and the program will run the reconstruction. Once it finishes,the reconstructed image is displayed on the screen at the bottom of the GUI. Theprogram also saves a DICOM file of the image in the location specified by theuser when loading the nuclear medicine data. This DICOM file can be opened byany external software capable of reading this type of file. At this point, the imagevoxels represent counts. In order to convert these counts into activity, a cameracalibration factor is required. As stated in the MIRD pamphlet number 23 [60],an experiment with a known source of activity should be done and two methodsare available; a point like source or a tomographic scan which approximates thepatient geometry. The GUI allows to input the information about the calibrationexperiment (i.e. time of measurement, activity, and isotope) and the normaliza-tion factor is determined accordingly (Fig. 7.7). A popup box opens asking if the1477.5. Running the Reconstruction and Determining the Normalization Factor of the CameraFigure 7.6: Example of a list of saved protocols with the details for the first one atthe bottom.1487.6. Testing the GUI and QuantificationFigure 7.7: Input information to determine the sensitivity or normalization factorfor the camera. In this case, an example of a sensitivity scan with 1000MBq of177Lu, performed at 12:15 on December 18 of 2014 is being shown. For this casein particular, a sensitivity factor of 0.6Counts/(min∗ kBq) was obtained.experiment was planar or tomographic and calculates the normalization or cali-bration factor for the camera. The sensitivity factor is given in Counts/(min*kBq)and is used to convert the image to activity values required for dosimetry.7.6 Testing the GUI and QuantificationThe SPEQToR GUI has been tested on the data obtained in experiments usingseveral isotopes including 99mTc, 111In, 123I, 131In, and now 177Lu [151]. Goodactivity quantification has been obtained for all of them.1497.7. ResultsFigure 7.8: Full view of SPEQToR.7.7 ResultsFigure 7.8 shows the complete screen shot of SPEQToR with all the sections men-tioned before.7.8 ConclusionsThe SPEQToR GUI provides the researcher with full control over the parametersof the reconstruction algorithm. Accuracy of quantification of different isotopesand phantoms have been tested using the methods implemented into this GUI. TheDICOM file of the reconstructed image is easily loaded in any dosimetry software1507.8. Conclusionssuch as JADA.151Chapter 8Conclusions8.1 Summary and FindingsThe thesis began with a short review of methods used to diagnose and treat Neu-roendocrine tumors (NETs). Among all the treatment methods, peptide receptorradionuclide therapy (PRRT) with somatostatin analogues labeled with 177Lu, hasshown the most promising results, improving quality of life and slowing tumorprogression.Accepted kidney toxicity levels which limit treatments with 177Lu have beenset in a range between 23 and 27 Gy. Presently however, a “one dose fits all”approach is used, in which every patient is injected with the same amount of ac-tivity (approximately 7400 MBq) in each treatment session. This approach is farfrom being optimal as it does not account for differences in organ uptakes be-tween patients. This has led to some patients being under-treated, while othersmay receive higher dose to the kidneys than the accepted limits. It is believedthat treatment plans using radioactivity injections based on an individualized ra-diation dose assessment could significantly improve PRRT outcomes and shouldbecome routine as is the case in external beam radiotherapies. To achieve this, ac-1528.1. Summary and Findingscurate estimates of doses delivered to healthy organs and tumors are required, butit has been wrongly believed that the task is very difficult and time consuming. Tobe able to determine the delivered dose, a series of nuclear medicine scans are re-quired, and quantitative corrections for physical effects that degrade image qualitymust be applied. This project was focused on developing a clinically feasibleprotocol for accurate quantification of 177Lu in order to perform personal-ized dosimetry calculations in the treatment of NETs. It was conducted atthe MIRG of the University of British Columbia in Vancouver.Chapter 3 analyzed the Bremsstrahlung (BRS) production caused by β parti-cles emitted by 177Lu in tissue, its contribution to the energy spectrum detected bya gamma camera, and the effect of using different collimators for imaging. Theresults of Monte-Carlo (MC) simulations suggested that the BRS contribution tothe 177Lu spectra recorded by the camera is minimal (less than 0.2%), and has nodegrading effects on image quantification. The use of medium energy collimatorsand an energy window centered on the 208 keV photopeak is recommended. Ad-ditional energy windows on each of the sides of this photopeak window should beset in order to correct for scatter using triple energy window (TEW) method.Chapter 4 studied the accuracy of quantification of this technique. For thispurpose, several phantom experiments with low count rates were performed. Im-age reconstructions were performed using the OSEM algorithm with attenuationand scatter correction. The determination of the camera normalization factor andthe quantification accuracy were studied using two different scatter correctionmethods; analytical photon distribution interpolated (APDI) and TEW. When1538.1. Summary and Findingsdetermining the camera normalization factor, our results showed that, as long asscatter correction is applied to planar scans the results are identical to normal-ization factors obtained by performing SPECT reconstructions. Therefore, it isrecommended that planar scans should be used for camera sensitivity determina-tion because they are not only easier to perform but are also less time consuming.For phantoms scanned in air, quantification within 5% and 10% was achievedwhen performing TEW and APDI scatter correction, respectively. For phantomsscanned in water, APDI scatter correction provided more accurate determinationof activity (within 5%). Also, our results suggest that APDI corrects better forscatter in challenging situations with non-uniform density of tissue. When back-ground activity was added, three segmentation methods were tested, 40% fixedthreshold, true volume of the objects, and our IADT method. In all cases theactivity was underestimated. However, based on the results of scans in air andwater, the problem does not seem to be related to the reconstruction algorithmbut rather to the segmentation method. IADT provided the best quantification ac-curacy which was very similar for TEW and APDI corrections. The differencesshown between the two scatter correction algorithms, are compensated by the cal-ibration curves of the IADT method. Although dosimetry for very small lesions isstill a challenge and better segmentation algorithms are required, IADT is a goodmethod when performing dosimetry for large organs such as the kidneys. Basedon these results, we believe that our protocol would underestimate the activity inkidneys within 5%. Finally, in cases where lesions are located in a medium withnon-uniform attenuation coefficient distribution, it is suggested to perform scatter1548.1. Summary and Findingscorrection using the APDI algorithm instead of TEW.In Chapter 5 camera deatime (DT) effects were studied. Correction for DT isimportant in therapy procedures as injected activities are large. Measurements fora Siemens SymbiaT and GE Hawkeye SPECT dual head cameras were performedby adding activity to a bottle placed off-center in a cylindrical phantom filled withwater. Deadtime values (τ) for the paralyzable model were determined for the fullspectrum and two photopeak windows centered at the 113 keV and 208 keV. Thechoice of the paralyzable model showed to be appropriate for the range of countrates studied.The count rate in the photopeak windows was calculated by removing scatterusing the TEW scatter correction method. Our results suggest that deadtime cor-rections should be performed based on scatter corrected photopeak window andnot using the full spectrum deadtime. Based on scatter and attenuation correction,it has been said that quantification of 177Lu is better achieved when the 208keVgamma peak is used for this purpose. Our study also shows that deadtime effectsare less important for this photopeak.Chapter 6 briefly described dosimetry calculations performed for 4 patients us-ing OLINDA, organ level, and voxelized dosimetry methods. The results showedthat dose values were the lowest when using OLINDA, but in all situations thedose delivered to the kidneys was lower than the suggested limit of 25Gy. More-over, the behavior of uptake in each organ was different. This strongly suggeststhat personalized dosimetry should be used.1558.2. Contribution to the Field8.2 Contribution to the FieldA quantitative protocol, which includes image reconstruction with attenuation,scatter, and deadtime corrections, has been developed and tested for use in per-sonalized dosimetry of patients undergoing peptide receptor radionuclide thera-pies. Several functions programmed in C++ were incorporated into a quantitativereconstruction graphical user interface (GUI), single photon emission quantitativetomographic reconstruction (SPEQToR), with the purpose of easily implementingour protocol and reconstruct data obtained with any manufacturer’s camera. Ithas been shown that personalized dosimetry is feasible and tools have been madeavailable for this purpose.8.3 Suggestions for Future WorkThis project was focused on the development of the protocol to determine theactivity distribution of 177Lu using SPECT/CT scans. A second step of the pro-cess necessary for accurate dosimetry involves the determination of the behaviorof the activity with time within the organs, and its combination with S-values.A method using a mono-exponential decay model and convolving pre-calculatedS-value kernels to estimate dose was presented. More studies are needed to deter-mine if the biokinetics is actually mono-exponential, or if it is better approximatedby a bi-exponential or a different function.The OLINDA type dosimetry assumes uniform distribution of activity usingpre-established phantoms with very specific geometries that do not represent indi-1568.3. Suggestions for Future Workvidual patients. A better method for dosimetry is when using the S-value kernelsthat take into account each patient’s geometry and activity distribution. In thisthesis we assumed both a uniform distribution of activity and a uniform washoutof activity to all the voxels in a desired region of interest. For the future, studiesthat investigate how accurate is this assumption should be performed to move intothe direction of finding exact voxel by voxel activity washout. To this end, anaccurate image registration is necessary.As the biggest limitation for quantification seemed to be related to the seg-mentation algorithms, more effort should be made in developing new segmenta-tion techniques for the nuclear medicine field that will facilitate the determinationof activity and volume of organs. This project dealt with dose to the kidney asthey constitute a limiting factor for the injected activity. However, bone marrowtoxicity is recognized as a limiting factor and techniques to estimate it should bedeveloped. All these questions will be addressed in future clinical trials performedin collaboration with centers in Ontario and BC. Correlations between doses andtreatment outcome will then be possible to investigate.157Bibliography[1] I. M. Modlin, K. Oberg, D. C. Chung, R. T. Jensen, W. W. D. Herder,R. V. Thakker, M. Caplin, G. D. Fave, G. A. Kaltsas, E. P. Krenning,S. F. Moss, O. Nilsson, G. Rindi, R. Salazar, P. Ruszniewski, A. Sundin,Gastroenteropancreatic neuroendocrine tumours, Lancet Oncol. (January)(2008) 61–72.[2] Siemens Medical Solutions, Symbia S and T, last accessed April 2015(2010).URL usa.healthcare.siemens.com/siemens_hwem-hwem_ssxa_websites-context-root/wcm/idc/groups/public/@us/@imaging/@molecular/documents/download/mda1/mdkw/~edisp/symbia-t-spec-sheet-2010-01977049.pdf[3] K. M. H, B. I. Al, B. Emily, B. Jordan, B. Lawrence, C. Michael, C. Orlo,D. Gerard, E. James, E. Lyska, E. Paul, G. Whitney, H. Martin, K. Fouad,K. Pamela, K. Boris, M. Jeffrey, P. Venu, S. Leonard, S. David, NCCN Clin-ical Practice Guidelines in Oncology: neuroendocrine tumors., NationalComprehensive Cancer Network : JNCCN 10 (6) (2012) 724–764.158Bibliography[4] G. a. Kaltsas, G. M. Besser, A. B. Grossman, The diagnosis and medi-cal management of advanced neuroendocrine tumors, Endocrine Reviews25 (March) (2004) 458–511. doi:10.1210/er.2003-0014.[5] J. C. Yao, M. Hassan, A. Phan, C. Dagohoy, C. Leary, J. E. Mares, E. K.Abdalla, J. B. Fleming, J. N. Vauthey, A. Rashid, D. B. Evans, One hundredyears after "carcinoid": Epidemiology of and prognostic factors for neu-roendocrine tumors in 35,825 cases in the United States, Journal of ClinicalOncology 26 (18) (2008) 3063–3072. doi:10.1200/JCO.2007.15.4377.[6] Y. Krausz, Z. Keidar, I. Kogan, E. Even-Sapir, R. Bar-Shalom, A. Engel,R. Rubinstein, J. Sachs, M. Bocher, S. Agranovicz, R. Chisin, O. Israel,SPECT/CT hybrid imaging with 111In-pentetreotide in assessment of neu-roendocrine tumours., Clinical endocrinology 59 (5) (2003) 565–73.URL http://www.ncbi.nlm.nih.gov/pubmed/14616879[7] S. Adams, R. Baum, T. Rink, S.-D. Petra-Maria, K.-h. Usadel, G. Hör,Original article Limited value of fluorine-18 fluorodeoxyglucose positronemission tomography for the imaging of neuroendocrine tumours, Euro-pean Journal of Nuclear Medicine 25 (1) (1998) 79–83.[8] K. P. Koopmans, E. G. de Vries, I. P. Kema, P. H. Elsinga, O. C. Neels, W. J.Sluiter, A. N. van der Horst-Schrivers, P. L. Jager, Staging of carcinoidtumours with 18F-DOPA PET: a prospective, diagnostic accuracy study,Lancet Oncology 7 (9) (2006) 728–734. doi:10.1016/S1470-2045(06)70801-4.159Bibliography[9] F. Montravers, D. Grahek, K. Kerrou, P. Ruszniewski, V. de Beco, N. Aide,F. Gutman, J.-D. Grangé, J.-P. Lotz, J.-N. Talbot, Can fluorodihydrox-yphenylalanine PET replace somatostatin receptor scintigraphy in patientswith digestive endocrine tumors?, Journal of nuclear medicine : officialpublication, Society of Nuclear Medicine 47 (9) (2006) 1455–1462.[10] E. P. Krenning, D. J. Kwekkeboom, R. Valkema, S. Pauwels, L. K. Kvols,M. De Jong, Peptide receptor radionuclide therapy, Annals of the New YorkAcademy of Sciences 1014 (1) (2004) 234–245. doi:10.1196/annals.1294.026.URL http://doi.wiley.com/10.1196/annals.1294.026[11] M. V. Essen, Peptide Receptor Radionuclide Therapy with 177 Lu-octreotate : Clinical Aspects, 2012.[12] M. Gabriel, C. Decristoforo, D. Kendler, G. Dobrozemsky, D. Heute,C. Uprimny, P. Kovacs, E. Von Guggenberg, R. Bale, I. J. Virgolini, 68Ga-DOTA-Tyr3-Octreotide PET in Neuroendocrine Tumors: Comparison withSomatostatin Receptor Scintigraphy and CT, Journal of Nuclear Medicine48 (4) (2007) 508–518. doi:10.2967/jnumed.106.035667.URL http://jnm.snmjournals.org/cgi/doi/10.2967/jnumed.106.035667[13] J. C. Reubi, B. Waser, J. C. Schaer, J. A. Laissue, Somatostatin receptorsst1-sst5 expression in normal and neoplastic human tissues using recep-160Bibliographytor autoradiography with subtype-selective ligands., European journal ofnuclear medicine 28 (2001) 836–846. doi:10.1007/s002590100598.[14] E. P. Krenning, D. J. Kwekkeboom, W. H. Bakker, W. a. P. Breeman,P. P. M. Kooij, H. Y. Oei, M. van Hagen, P. T. E. Postema, M. de Jong,J. C. Reubi, T. J. Visser, a. E. M. Reijs, L. J. Hofland, J. W. Koper, S. W. J.Lamberts, Somatostatin receptor scintigraphy with [111In-DTPA-d-Phe1]-and [123I-Tyr3]-octreotide: the Rotterdam experience with more than 1000patients, European Journal of Nuclear Medicine 20 (8) (1993) 716–731.doi:10.1007/BF00181765.[15] E. Seregni, A. Chiti, E. Bombardieri, Radionuclide imaging of neu-roendocrine tumours: Biological basis and diagnostic results, EuropeanJournal of Nuclear Medicine 25 (6) (1998) 639–658. doi:10.1007/s002590050267.[16] C. Decristoforo, S. J. Mather, W. Cholewinski, E. Donnemiller, G. Ric-cabona, R. Moncayo, (99m)Tc-EDDA/HYNIC-TOC: A new (99m)Tc-labelled radiopharmaceutical for imaging somatostatin receptor-positivetumours: First clinical results and intra-patient comparison with 111In-labelled octreotide derivatives, European Journal of Nuclear Medicine27 (9) (2000) 1318–1325. doi:10.1007/s002590000289.[17] A. Plachcinska, R. Mikolajczak, H. R. Maecke, E. Mlodkowska, J. Kunert-Radek, A. Michalski, K. Rzeszutek, J. Kozak, J. Kusmierek, Clinical161Bibliographyusefulness of 99mTc-EDDA/HYNIC-TOC scintigraphy in oncological di-agnostics: A preliminary communication, European Journal of NuclearMedicine and Molecular Imaging 30 (10) (2003) 1402–1406. doi:10.1007/s00259-003-1254-6.[18] C. Rafal, P. M. Gemma, K. Jerzy, M. Reanata, Z. Katarzyna, G. Maria,S. Jerzy, S. Alberto, Somatostatin receptor scintigraphy using 99mTc-EDDA/HYNIC-TOC in patients with medullary thyroid carcinoma, Euro-pean Journal of Nuclear Medicine and Molecular Imaging 34 (2007) 1635–1645. doi:10.1007/s00259-007-0479-1.[19] J. Grimes, A. Celler, B. Birkenfeld, S. Shcherbinin, M. H. Listewnik,H. Piwowarska-Bilska, R. Mikolajczak, P. Zorga, Patient-specific radia-tion dosimetry of 99mTc-HYNIC-Tyr3-octreotide in neuroendocrine tu-mors., Journal of nuclear medicine : official publication, Society of NuclearMedicine 52 (9) (2011) 1474–81. doi:10.2967/jnumed.111.088203.URL http://www.ncbi.nlm.nih.gov/pubmed/21795364[20] J.-M. Beauregard, M. S. Hofman, G. Kong, R. J. Hicks, The tumoursink effect on the biodistribution of 68Ga-DOTA-octreotate: implicationsfor peptide receptor radionuclide therapy., European journal of nuclearmedicine and molecular imaging 39 (1) (2012) 50–6. doi:10.1007/s00259-011-1937-3.URL http://www.ncbi.nlm.nih.gov/pubmed/21932117[21] J. K. Ramage, A. Ahmed, J. Ardill, N. Bax, D. J. Breen, M. E. Caplin,162BibliographyP. Corrie, J. Davar, A. H. Davies, V. Lewington, T. Meyer, J. Newell-Price,G. Poston, N. Reed, A. Rockall, W. Steward, R. V. Thakker, C. Toubanakis,J. Valle, C. Verberke, A. B. Grossman, Guidelines for the managementof gastroenteropancreatic neuroendocrine (including carcinoid) tumours(NETs)., Gut 61 (2012) 6–32. doi:10.1136/gutjnl-2011-300831.[22] I. M. Modlin, M. Kidd, I. Latich, M. N. Zikusoka, M. D. Shapiro, Cur-rent status of gastrointestinal carcinoids, Gastroenterology 128 (6) (2005)1717–1751. doi:10.1053/j.gastro.2005.03.038.[23] J. a. Norton, Surgery for primary pancreatic neuroendocrine tumors., Jour-nal of gastrointestinal surgery : official journal of the Society for Surgeryof the Alimentary Tract 10 (3) (2006) 327–331. doi:10.1016/j.gassur.2005.08.023.[24] F. G. I. van Vilsteren, E. S. Baskin-Bey, D. M. Nagorney, S. O. Sanderson,W. K. Kremers, C. B. Rosen, G. J. Gores, T. J. Hobday, Liver transplanta-tion for gastroenteropancreatic neuroendocrine cancers: Defining selectioncriteria to improve survival, Liver Transplantation 12 (3) (2006) 448–456.doi:10.1002/lt.[25] A. Frilling, M. Malago, F. Weber, A. Paul, S. Nadalin, G. C. Sotiropou-los, V. Cicinnati, S. Beckebaum, A. Bockisch, J. Mueller-Brand, M. Hof-mann, K. W. Schmid, G. Gerken, C. E. Broelsch, Liver Transplanta-tion for Patients With Metastatic Endocrine Tumors: Single-Center Ex-163Bibliographyperience With 15 Patients, Liver Transplantation 12 (2006) 1089–1096.doi:10.1002/lt.[26] M. Olausson, S. Friman, G. Herlenius, C. Cahlin, O. Nilsson, S. Jansson,B. Wängberg, H. k. Ahlman, Orthotopic Liver of Multivisceral Transplan-tation as Treatment of Metastatic Neuroendocrine Tumo, Liver Transplan-tation 13 (3) (2007) 327–333. doi:10.1002/lt.[27] Y. P. Le Treut, E. Grégoire, J. Belghiti, O. Boillot, O. Soubrane, G. Man-tion, D. Cherqui, D. Castaing, P. Ruszniewski, P. Wolf, F. Paye, E. Salame,F. Muscari, F. R. Pruvot, J. Baulieux, Predictors of long-term survival afterliver transplantation for metastatic endocrine tumors: An 85-case Frenchmulticentric report, American Journal of Transplantation 8 (6) (2008)1205–1213. doi:10.1111/j.1600-6143.2008.02233.x.[28] D. O’Toole, M. Ducreux, G. Bommelaer, J. L. Wemeau, O. Bouché,F. Catus, J. Blumberg, P. Ruszniewski, Treatment of carcinoid syn-drome: A prospective crossover evaluation of lanreotide versus oc-treotide in terms of efficacy, patient acceptability, and tolerance, Cancer88 (4) (2000) 770–776. doi:10.1002/(SICI)1097-0142(20000215)88:4<770::AID-CNCR6>3.0.CO;2-0.[29] A. Rinke, H. H. Müller, C. Schade-Brittinger, K. J. Klose, P. Barth,M. Wied, C. Mayer, B. Aminossadati, U. F. Pape, M. Bläker, J. Harder,C. Arnold, T. Gress, R. Arnold, Placebo-controlled, double-blind, prospec-tive, randomized study on the effect of octreotide LAR in the control of164Bibliographytumor growth in patients with metastatic neuroendocrine midgut tumors: Areport from the PROMID study group, Journal of Clinical Oncology 27 (28)(2009) 4656–4663. doi:10.1200/JCO.2009.22.8510.[30] K. Oberg, Interferon in the management of neuroendocrine GEP-tumors: areview., Digestion 62 Suppl 1 (2000) 92–97. doi:10.1159/000051862.[31] L. Kölby, G. Persson, S. Franzén, B. Ahrén, Randomized clinical trial of theeffect of interferon alpha on survival in patients with disseminated midgutcarcinoid tumours., The British journal of surgery 90 (6) (2003) 687–693.doi:10.1002/bjs.4149.[32] M. K. Krazyzanowska, M. S. Tsao, a. M. Oza, M. Haider, R. Feld, J. Knox,S. Chin, H. Hu, L. L. Siu, Capecitabine plus rofecoxib show no activityin patients with metastatic neuroendocrine tumours [6], Clinical Oncology18 (1) (2006) 88–89. doi:10.1016/j.clon.2005.08.012.[33] E. Bajetta, L. Catena, G. Procopio, S. De Dosso, E. Bichisao, L. Fer-rari, A. Martinetti, M. Platania, E. Verzoni, B. Formisano, R. Ba-jetta, Are capecitabine and oxaliplatin (XELOX) suitable treatments forprogressing low-grade and high-grade neuroendocrine tumours?, CancerChemotherapy and Pharmacology 59 (5) (2007) 637–642. doi:10.1007/s00280-006-0306-6.[34] J. R. Strosberg, R. L. Fine, J. Choi, A. Nasir, D. Coppola, D. T. Chen,J. Helm, L. Kvols, First-line chemotherapy with capecitabine and temo-165Bibliographyzolomide in patients with metastatic pancreatic endocrine carcinomas, Can-cer 117 (2) (2011) 268–275. doi:10.1002/cncr.25425.[35] L. K. Kvols, M. Buck, Chemotherapy of metastatic carcinoid and islet celltumors. A review., The American journal of medicine 82 (5B) (1987) 77–83.[36] I. Duran, J. Kortmansky, D. Singh, H. Hirte, W. Kocha, G. Goss, L. Le,a. Oza, T. Nicklee, J. Ho, D. Birle, G. R. Pond, D. Arboine, J. Dancey,S. Aviel-Ronen, M.-S. Tsao, D. Hedley, L. L. Siu, A phase II clini-cal and pharmacodynamic study of temsirolimus in advanced neuroen-docrine carcinomas., British journal of cancer 95 (9) (2006) 1148–1154.doi:10.1038/sj.bjc.6603419.[37] K. Sxhupak, K. Wallner, The role of radiation therapy in the treatment oflocally unresectable or metastatic carcinoid tumors, International Journalof Radiation Oncology 20 (1991) (1991) 489–495.URL http://www.sciencedirect.com/science/article/pii/0360301691900618[38] M. van Essen, E. P. Krenning, B. L. R. Kam, M. de Jong, R. Valkema,D. J. Kwekkeboom, Peptide-receptor radionuclide therapy for endocrinetumors., Nature reviews. Endocrinology 5 (7) (2009) 382–393. doi:10.1038/nrendo.2009.105.URL http://dx.doi.org/10.1038/nrendo.2009.105166Bibliography[39] L. Bodei, M. Cremonesi, M. Ferrari, M. Pacifici, C. M. Grana, M. Bar-tolomei, S. M. Baio, M. Sansovini, G. Paganelli, Long-term evaluationof renal toxicity after peptide receptor radionuclide therapy with 90Y-DOTATOC and 177Lu-DOTATATE: The role of associated risk factors, Eu-ropean Journal of Nuclear Medicine and Molecular Imaging 35 (10) (2008)1847–1856. doi:10.1007/s00259-008-0778-1.[40] R. B. Christiane Schuchardt, Harshad R. Kulkarni, Vikas Prasad, CarolinZachert, Dirk Müller, The Bad Berka Dose Protocol: ComparativeResults of Dosimetry in Peptide Receptor Radionuclide Therapy Us-ing 177Lu-DOTATATE, 177LU-DOTANOC, and 177Lu-DOTATOC,in: Recent Results in Cancer Research, Vol. 194, 2013, pp. 301–317.doi:10.1007/978-3-642-27994-2.URL http://www.springerlink.com/index/10.1007/978-3-642-27994-2[41] J. P. Esser, E. P. Krenning, J. J. M. Teunissen, P. P. M. Kooij, a. L. H.Van Gameren, W. H. Bakker, D. J. Kwekkeboom, Comparison of [177Lu-DOTA0,Tyr3]octreotate and [177Lu-DOTA0,Tyr3]octreotide: Which pep-tide is preferable for PRRT?, European Journal of Nuclear Medicineand Molecular Imaging 33 (11) (2006) 1346–1351. doi:10.1007/s00259-006-0172-9.[42] B. L. R. Kam, J. J. M. Teunissen, E. P. Krenning, W. W. De Herder, S. Khan,E. I. Van Vliet, D. J. Kwekkeboom, Lutetium-labelled peptides for ther-167Bibliographyapy of neuroendocrine tumours, European Journal of Nuclear Medicine andMolecular Imaging 39. doi:10.1007/s00259-011-2039-y.[43] T. D. Poeppel, I. Binse, S. Petersenn, H. Lahner, M. Schott, G. An-toch, W. Brandau, A. Bockisch, C. Boy, 68Ga-DOTATOC Versus 68Ga-DOTATATE PET/CT in Functional Imaging of Neuroendocrine Tumors,Journal of Nuclear Medicine 52 (12) (2011) 1864–1870. doi:10.2967/jnumed.111.091165.[44] I. Velikyan, A. Sundin, J. Sörensen, M. Lubberink, M. Sandström,U. Garske-Román, H. Lundqvist, D. Granberg, B. Eriksson, Quantitativeand Qualitative Intrapatient Comparison of 68Ga-DOTATOC and 68Ga-DOTATATE: Net Uptake Rate for Accurate Quantification., Journal of nu-clear medicine : official publication, Society of Nuclear Medicine 55 (2)(2014) 204–10. doi:10.2967/jnumed.113.126177.URL http://www.ncbi.nlm.nih.gov/pubmed/24379222[45] D. Wild, J. B. Bomanji, P. Benkert, H. Maecke, P. J. Ell, J. C. Reubi, M. E.Caplin, Comparison of 68Ga-DOTANOC and 68Ga-DOTATATE PET/CTwithin patients with gastroenteropancreatic neuroendocrine tumors., Jour-nal of nuclear medicine : official publication, Society of Nuclear Medicine54 (3) (2013) 364–72. doi:10.2967/jnumed.112.111724.URL http://www.ncbi.nlm.nih.gov/pubmed/23297077[46] L. B. Anthony, E. a. Woltering, G. D. Espenan, M. D. Cronin, T. J. Maloney,K. E. McCarthy, Indium-111-pentetreotide prolongs survival in gastroen-168Bibliographyteropancreatic malignancies., Seminars in nuclear medicine 32 (2) (2002)123–132. doi:10.1053/snuc.2002.31769.[47] R. Valkema, M. De Jong, W. H. Bakker, W. a. P. Breeman, P. P. M. Kooij,P. J. Lugtenburg, F. H. De Jong, A. Christiansen, B. L. R. Kam, W. W. DeHerder, M. Stridsberg, J. Lindemans, G. Ensing, E. P. Krenning, Phase Istudy of peptide receptor radionuclide therapy with [In-DTPA]octreotide:the Rotterdam experience., Seminars in nuclear medicine 32 (2) (2002)110–122. doi:S0001299802500155[pii].[48] a. Otte, R. Herrmann, a. Heppeler, M. Behe, E. Jermann, P. Powell, H. R.Maecke, J. Muller, Yttrium-90 DOTATOC: First clinical results, EuropeanJournal of Nuclear Medicine 26 (11) (1999) 1439–1447. doi:10.1007/s002590050476.[49] C. Waldherr, M. Pless, H. Maecke, A. Haldemann, J. Mueller-Brand, The clinical value of [90Y-DOTAl-D-Phe1Tyr3octreotide (90Y-DOTATOC) in the treatment of neuroendocrine tumours: A clinicalphase II study, Ann. Oncol. 12 (3) (2001) 941–945. doi:10.1097/00000658-191909000-00006.[50] C. Waldherr, M. Pless, H. R. Maecke, T. Schumacher, A. Crazzolara, E. U.Nitzsche, A. Haldemann, J. Mueller-brand, Tumor Response and ClinicalBenefit in Neuroendocrine Tumors After 7.4 GBq 90Y-DOTATOC, TheJournal of Nuclear Medicine 43 (5) (2002) 610–616.169Bibliography[51] M. Chinol, L. Bodei, M. Cremonesi, G. Paganelli, Receptor-mediated ra-diotherapy with Y-DOTA-DPhe-Tyr-octreotide: the experience of the Eu-ropean Institute of Oncology Group., Seminars in nuclear medicine 32 (2)(2002) 141–147. doi:S0001299802500180[pii].[52] D. Bushnell, T. O’Dorisio, Y. Menda, T. Carlisle, P. Zehr, M. Connolly,M. Karwal, S. Miller, S. Parker, H. Bouterfa, Evaluating the clinical effec-tiveness of 90Y-SMT 487 in patients with neuroendocrine tumors., Journalof nuclear medicine : official publication, Society of Nuclear Medicine44 (10) (2003) 1556–1560.[53] R. Valkema, S. Pauwels, L. K. Kvols, R. Barone, F. Jamar, W. H.Bakker, D. J. Kwekkeboom, H. Bouterfa, E. P. Krenning, Survivaland response after peptide receptor radionuclide therapy with [90Y-DOTA0,Tyr3]octreotide in patients with advanced gastroenteropancreaticneuroendocrine tumors, Seminars in Nuclear Medicine 36 (2) (2006) 147–156. doi:10.1053/j.semnuclmed.2006.01.001.[54] D. J. Kwekkeboom, J. J. Teunissen, W. H. Bakker, P. P. Kooij, W. W. DeHerder, R. a. Feelders, C. H. Van Eijck, J. P. Esser, B. L. Kam, E. P. Kren-ning, Radiolabeled somatostatin analog [177Lu-DOTA0, Tyr3]octreotate inpatients with endocrine gastroenteropancreatic tumors, Journal of ClinicalOncology 23 (12) (2005) 2754–2762. doi:10.1200/JCO.2005.08.066.[55] D. J. Kwekkeboom, W. W. de Herder, C. H. J. van Eijck, B. L. Kam, M. vanEssen, J. J. M. Teunissen, E. P. Krenning, Peptide Receptor Radionuclide170BibliographyTherapy in Patients With Gastroenteropancreatic Neuroendocrine Tumors,Seminars in Nuclear Medicine 40 (2) (2010) 78–88. doi:10.1053/j.semnuclmed.2009.10.004.URL http://dx.doi.org/10.1053/j.semnuclmed.2009.10.004[56] J. J. M. Teunissen, D. J. Kwekkeboom, E. P. Krenning, Quality oflife in patients with gastroenteropancreatic tumors treated with [177Lu-DOTA0,Tyr3]octreotate, Journal of Clinical Oncology 22 (13) (2004)2724–2729. doi:10.1200/JCO.2004.10.016.[57] M. Garkavij, M. Nickel, K. Sjögreen-Gleisner, M. Ljungberg, T. Ohlsson,K. Wingå rdh, S. E. Strand, J. Tennvall, 177Lu-[DOTA0,Tyr3] octreotatetherapy in patients with disseminated neuroendocrine tumors: Analysis ofdosimetry with impact on future therapeutic strategy, Cancer 116 (2010)1084–1092. doi:10.1002/cncr.24796.[58] M. Stabin, The case for patient-specific dosimetry in radionuclide therapy,Cancer Biotherapy & Radiopharmaceuticals 23 (3).URL http://online.liebertpub.com/doi/abs/10.1089/cbr.2007.0445[59] J. Grimes, C. Uribe, A. Celler, JADA: A graphical user interface forcomprehensive internal dose assessment in nuclear medicine, MedicalPhysics 40 (7) (2013) 072501. doi:10.1118/1.4810963.URL http://link.aip.org/link/MPHYA6/v40/i7/p072501/s1&Agg=doi171Bibliography[60] Y. K. Dewaraja, E. C. Frey, G. Sgouros, A. B. Brill, P. Roberson, P. B.Zanzonico, M. Ljungberg, MIRD Pamphlet No. 23: Quantitative SPECTfor Patient-Specific 3-Dimensional Dosimetry in Internal RadionuclideTherapy, Journal of Nuclear Medicine 53 (8) (2012) 1310–1325. doi:10.2967/jnumed.111.100123.URL http://www.ncbi.nlm.nih.gov/pubmed/22743252[61] S. R. Cherry, J. Sorenson, M. E. Phelps, B. M. Methe, Physics in Nu-clear Medicine, forth edit Edition, Vol. 31, Elsevier Saunders, Philadelphia,2004. doi:10.1118/1.1776595.[62] P. Kozlowski, UBC PHYS-540: Radiological Imaging, Lecture Notes(2010).[63] S. Bacca, L. Doria, S. Yen, UBC PHYS-505: Nuclear Physics, LectureNotes (2012).[64] A. Celler, UBC PHYS-541: Nuclear Medicine, Lecture Notes (2012).[65] F. G. Kondev, Nuclear Data Sheets for A = 177, Nuclear Data Sheets98 (03) (2003) 801–1095. doi:10.1006/ndsh.2003.0006.URL http://linkinghub.elsevier.com/retrieve/pii/S0090375203900068[66] M. Cremonesi, M. Ferrari, L. Bodei, G. Tosi, G. Paganelli, Dosimetryin Peptide radionuclide receptor therapy: a review., Journal of nuclear172Bibliographymedicine : official publication, Society of Nuclear Medicine 47 (2006)1467–1475.[67] R. Mikolajczak, P. Jl, Reactor Produced Beta-emitting Nuclides for NuclearMedicine, World journal of nuclear medicine 36 (44) (2005) 184–190.[68] Iaea, Manual for reactor produced radioisotopes (January) (2003) 1–254.URL http://www.isotopes.gov/outreach/reports/Reactor_Isotopes.pdf[69] M. R. a. Pillai, S. Chakraborty, T. Das, M. Venkatesh, N. Ramamoorthy,Production logistics of 177Lu for radionuclide therapy, Applied Radiationand Isotopes 59 (2-3) (2003) 109–118. doi:10.1016/S0969-8043(03)00158-1.[70] B. Singh, Nuclear data sheets for A = 188 (2002). doi:10.1016/j.nds.2004.01.002.[71] E. Browne, Nuclear Data Sheets for A = 90 *, Nuclear Data Sheets 82 (3)(1997) 379–546.[72] J. N. Aarsvold, Emission Tomography The Fundamentals of PET andSPECT, Elsevier Academic Press, 2004.[73] P. P. Bruyant, Analytic and iterative reconstruction algorithms in SPECT.,Journal of nuclear medicine : official publication, Society of NuclearMedicine 43 (10) (2002) 1343–58.URL http://www.ncbi.nlm.nih.gov/pubmed/12368373173Bibliography[74] H. M. Hudson, R. S. Larkin, Accelerated image reconstruction using or-dered subsets of projection data., IEEE transactions on medical imaging13 (4) (1994) 601–609. doi:10.1109/42.363108.URL http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=363108http://www.ncbi.nlm.nih.gov/pubmed/18218538[75] L. Shepp, Y. Vardi, Maximum likelihood reconstruction for emission to-mography, Medical Imaging, IEEE Transactions on 1 (2) (1982) 113–122.URL http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=4307558[76] F. Beekman, Efficient fully 3-D iterative SPECT reconstruction withMonte Carlo-based scatter compensation, Medical Imaging, IEEE. . . 21 (8) (2002) 867–77. doi:10.1109/TMI.2002.803130.URL http://www.ncbi.nlm.nih.gov/pubmed/12472260http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=1076032[77] I. Buvat, M. Rodriguez-Villafuerte, a. Todd-Pokropek, H. Benali, R. DiPaola, Comparative assessment of nine scatter correction methods basedon spectral analysis using Monte Carlo simulations., Journal of nuclearmedicine : official publication, Society of Nuclear Medicine 36 (8) (1995)1476–1488.[78] M. Ljungberg, M. a. King, G. J. Hademenos, S. E. Strand, Comparison offour scatter correction methods using Monte Carlo simulated source dis-174Bibliographytributions., Journal of nuclear medicine : official publication, Society ofNuclear Medicine 35 (1) (1994) 143–151.[79] K. Ogawa, Y. Harata, T. Ichihara, A. Kubo, S. Hashimoto, A practicalmethod for position-dependent Compton-scatter correction in single photonemission CT., IEEE transactions on medical imaging 10 (3) (1991) 408–12.doi:10.1109/42.97591.URL http://www.ncbi.nlm.nih.gov/pubmed/18222843[80] E. Vandervoort, A. Celler, G. Wells, S. Blinder, K. Dixon, S. Member,Y. Pang, Implementation of an Analytically Based Scatter Correction inSPECT Reconstructions, IEEE Transactions on Nuclear Science 52 (3)(2005) 645–653.[81] R. G. Wells, a. Celler, R. Harrop, Analytical calculation of photon distribu-tions in SPECT projections, IEEE Transactions on Nuclear Science 45 (6)(1998) 3202–3214. doi:10.1109/23.736199.URL http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=736199[82] C. A. Quarles, Bremsstrahlung: An experimentalist’s personal perspectiveon the postmodern era, Radiation Physics and Chemistry 59 (2) (2000)159–169. doi:10.1016/S0969-806X(00)00287-5.URL http://linkinghub.elsevier.com/retrieve/pii/S0969806X00002875175Bibliography[83] Shivaramu, Spectral shapes for atomic-field bremsstrahlung produced bybeta particles of yttrium-90 in thick targets of Cu. Cd, Ta and Pb, Journalof Physics B: Atomic and Molecular Physics 19 (13) (1986) 1935–1944.doi:10.1088/0022-3700/19/13/007.URL http://stacks.iop.org/0022-3700/19/i=13/a=007?key=crossref.d7a9457d927820d84adbb3d5c82bdce6[84] G. R. Blumenthal, R. J. Gould, Bremsstrahlung, Synchrotron Radiation,and Compton Scattering of High Energy Electrons Traversing Dilute Gases,Reviews of Modern Physics 42 (2) (1970) 237–271.URL http://rmp.aps.org/abstract/RMP/v42/i2/p237_1[85] T. Carlier, T. Eugène, C. Bodet-Milin, E. Garin, C. Ansquer, C. Rousseau,L. Ferrer, J. Barbet, F. Schoenahl, F. Kraeber-Bodéré, Assessment of ac-quisition protocols for routine imaging of Y-90 using PET/CT., EJNMMIresearch 3 (1) (2013) 11. doi:10.1186/2191-219X-3-11.URL http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3614476&tool=pmcentrez&rendertype=abstract[86] L. P. Clarke, S. J. Cullom, R. Shaw, C. Reece, B. C. Penney, M. A. King,M. Silbiger, Bremsstrahlung imaging using the gamma camera: factors af-fecting attenuation., Journal of nuclear medicine : official publication, So-ciety of Nuclear Medicine 33 (1) (1992) 161–166.URL http://www.ncbi.nlm.nih.gov/pubmed/1730984[87] D. Minarik, M. Ljungberg, P. Segars, K. S. Gleisner, Evaluation of quantita-176Bibliographytive planar 90Y bremsstrahlung whole-body imaging., Physics in medicineand biology 54 (19) (2009) 5873–83. doi:10.1088/0031-9155/54/19/014.URL http://www.ncbi.nlm.nih.gov/pubmed/19759410[88] S. Ito, H. Kurosawa, H. Kasahara, S. Teraoka, E. Ariga, S. Deji, M. Hirota,T. Saze, T. Minamizawa, K. Nishizawa, (90)Y bremsstrahlung emissioncomputed tomography using gamma cameras., Annals of nuclear medicine23 (3) (2009) 257–67. doi:10.1007/s12149-009-0233-9.URL http://www.ncbi.nlm.nih.gov/pubmed/19326187[89] D. Minarik, K. Sjögreen Gleisner, M. Ljungberg, Evaluation of quan-titative (90)Y SPECT based on experimental phantom studies., Physicsin medicine and biology 53 (20) (2008) 5689–703. doi:10.1088/0031-9155/53/20/008.URL http://www.ncbi.nlm.nih.gov/pubmed/18812648[90] X. Rong, Y. Du, M. Ljungberg, E. Rault, S. Vandenberghe, E. C.Frey, Development and evaluation of an improved quantitative (90)Ybremsstrahlung SPECT method., Medical physics 39 (5) (2012) 2346–58.doi:10.1118/1.3700174.URL http://www.ncbi.nlm.nih.gov/pubmed/22559605[91] M. Elschot, M. G. E. H. Lam, M. A. A. J. van den Bosch, M. A. Viergever,H. W. A. M. de Jong, Quantitative Monte Carlo-based 90Y SPECT recon-struction., Journal of nuclear medicine : official publication, Society of177BibliographyNuclear Medicine 54 (9) (2013) 1557–63. doi:10.2967/jnumed.112.119131.URL http://www.ncbi.nlm.nih.gov/pubmed/23907758[92] S. Heard, G. D. Flux, M. J. Guy, R. J. Ott, Monte Carlo simulation of 90Y Bremsstrahlung imaging, Conference Record, 2004 00 (C) (2004) 3579–3583.URL http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=1466658[93] X. Rong, Y. Du, E. C. Frey, A method for energy window opti-mization for quantitative tasks that includes the effects of model-mismatch on bias: application to Y-90 bremsstrahlung SPECTimaging., Physics in medicine and biology 57 (12) (2012) 3711–25.doi:10.1088/0031-9155/57/12/3711.URL http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3739432&tool=pmcentrez&rendertype=abstract[94] D. Autret, A. Bitar, L. Ferrer, A. Lisbona, M. Bardiès, Monte Carlomodeling of gamma cameras for I-131 imaging in targeted radiotherapy.,Cancer biotherapy & radiopharmaceuticals 20 (1) (2005) 77–84. doi:10.1089/cbr.2005.20.77.URL http://www.ncbi.nlm.nih.gov/pubmed/15778585[95] M. Ljungberg, Monte Carlo simulations for therapy imaging, Journalof Physics: Conference Series 317 (2011) 012016. doi:10.1088/178Bibliography1742-6596/317/1/012016.URL http://stacks.iop.org/1742-6596/317/i=1/a=012016?key=crossref.507a408df2ff44916b525cd3be2acb73[96] M. Pacilio, N. Lanconelli, S. Lo Meo, M. Betti, L. Montani, L. A. TorresAroche, M. A. Coca Pérez, Differences among Monte Carlo codes inthe calculations of voxel S values for radionuclide targeted therapy andanalysis of their impact on absorbed dose evaluations, Medical Physics36 (5) (2009) 1543. doi:10.1118/1.3103401.URL http://link.aip.org/link/MPHYA6/v36/i5/p1543/s1&Agg=doi[97] E. Rault, S. Staelens, R. Van Holen, J. De Beenhouwer, S. Vandenberghe,Fast simulation of yttrium-90 bremsstrahlung photons with GATE, MedicalPhysics 37 (6) (2010) 2943. doi:10.1118/1.3431998.URL http://link.aip.org/link/MPHYA6/v37/i6/p2943/s1&Agg=doi[98] X-5 Monte Carlo Team, MCNP5, LA-UR-03-1987.[99] S. Jan, G. Santin, D. Strul, S. Staelens, K. Assié, D. Autret, S. Avner,R. Barbier, M. Bardiès, P. M. Bloomfield, D. Brasse, V. Breton, P. Bruyn-donckx, I. Buvat, a. F. Chatziioannou, Y. Choi, Y. H. Chung, C. Comtat,D. Donnarieix, L. Ferrer, S. J. Glick, C. J. Groiselle, D. Guez, P.-F.Honore, S. Kerhoas-Cavata, a. S. Kirov, V. Kohli, M. Koole, M. Krieguer,179BibliographyD. J. V. D. Laan, F. Lamare, G. Largeron, C. Lartizien, D. Lazaro,M. C. Maas, L. Maigne, F. Mayet, F. Melot, C. Merheb, E. Pennacchio,J. Perez, U. Pietrzyk, F. R. Rannou, M. Rey, D. R. Schaart, C. R.Schmidtlein, L. Simon, T. Y. Song, J.-M. Vieira, D. Visvikis, R. V. D.Walle, E. Wieërs, C. Morel, GATE: a simulation toolkit for PET andSPECT, Physics in Medicine and Biology 49 (19) (2004) 4543–4561.doi:10.1088/0031-9155/49/19/007.URL http://stacks.iop.org/0031-9155/49/i=19/a=007?key=crossref.be1d0bbaf1d502515584e510c77393cb[100] M. G. Stabin, J. A. Siegel, Physical models and dose factors for use ininternal dose assessment., Health physics 85 (2003) 294–310. doi:10.1097/00004032-200309000-00006.[101] M. R. Bhat, Evalutated Nuclear Structure Data File (ENSDF), in: S. Qaim(Ed.), Nuclear Data for Science and Technology, Springer Berlin Heidel-berg, 1992, pp. 817–821. doi:10.1007/978-3-642-58113-7\_227.URL http://dx.doi.org/10.1007/978-3-642-58113-7_227[102] S. Agostinelli, J. Allison, K. Amako, J. Apostolakis, H. Araujo, P. Arce,M. Asai, D. Axen, S. Banerjee, G. Barrand, F. Behner, L. Bellagamba,J. Boudreau, L. Broglia, A. Brunengo, H. Burkhardt, S. Chauvie,J. Chuma, R. Chytracek, G. Cooperman, G. Cosmo, P. Degtyarenko,A. Dell’Acqua, G. Depaola, D. Dietrich, R. Enami, A. Feliciello,C. Ferguson, H. Fesefeldt, G. Folger, F. Foppiano, A. Forti, S. Garelli,180BibliographyS. Giani, R. Giannitrapani, D. Gibin, J. Gómez Cadenas, I. González,G. Gracia Abril, G. Greeniaus, W. Greiner, V. Grichine, A. Grossheim,S. Guatelli, P. Gumplinger, R. Hamatsu, K. Hashimoto, H. Hasui,A. Heikkinen, A. Howard, V. Ivanchenko, A. Johnson, F. Jones, J. Kallen-bach, N. Kanaya, M. Kawabata, Y. Kawabata, M. Kawaguti, S. Kelner,P. Kent, A. Kimura, T. Kodama, R. Kokoulin, M. Kossov, H. Kurashige,E. Lamanna, T. Lampén, V. Lara, V. Lefebure, F. Lei, M. Liendl, W. Lock-man, F. Longo, S. Magni, M. Maire, E. Medernach, K. Minamimoto,P. Mora de Freitas, Y. Morita, K. Murakami, M. Nagamatu, R. Nartallo,P. Nieminen, T. Nishimura, K. Ohtsubo, M. Okamura, S. O’Neale,Y. Oohata, K. Paech, J. Perl, A. Pfeiffer, M. Pia, F. Ranjard, A. Rybin,S. Sadilov, E. Di Salvo, G. Santin, T. Sasaki, N. Savvas, Y. Sawada,S. Scherer, S. Sei, V. Sirotenko, D. Smith, N. Starkov, H. Stoecker,J. Sulkimo, M. Takahata, S. Tanaka, E. Tcherniaev, E. Safai Tehrani,M. Tropeano, P. Truscott, H. Uno, L. Urban, P. Urban, M. Verderi,A. Walkden, W. Wander, H. Weber, J. Wellisch, T. Wenaus, D. Williams,D. Wright, T. Yamada, H. Yoshida, D. Zschiesche, Geant4-a simulationtoolkit, Nuclear Instruments and Methods in Physics Research Section A:Accelerators, Spectrometers, Detectors and Associated Equipment 506 (3)(2003) 250–303. doi:10.1016/S0168-9002(03)01368-8.URL http://linkinghub.elsevier.com/retrieve/pii/S0168900203013688[103] S. L. Meo, P. Bennati, M. N. Cinti, N. Lanconelli, F. L. Navarria,181BibliographyR. Pani, R. Pellegrini, A. Perrotta, F. Vittorini, A Geant4 simu-lation code for simulating optical photons in SPECT scintillationdetectors, Journal of Instrumentation 4 (07) (2009) P07002–P07002.doi:10.1088/1748-0221/4/07/P07002.URL http://stacks.iop.org/1748-0221/4/i=07/a=P07002?key=crossref.222fdbe9a88766bcdcc3536ac81ca390[104] E. Rault, S. Staelens, R. Van Holen, J. De Beenhouwer, S. Vanden-berghe, Accurate Monte Carlo modelling of the back compartments ofSPECT cameras., Physics in medicine and biology 56 (2011) 87–104.doi:10.1088/0031-9155/56/1/006.[105] K. Assié, I. Gardin, P. Véra, I. Buvat, Validation of the Monte Carlo sim-ulator GATE for indium-111 imaging., Physics in medicine and biology50 (13) (2005) 3113–25. doi:10.1088/0031-9155/50/13/010.URL http://www.ncbi.nlm.nih.gov/pubmed/15972984[106] S. Shcherbinin, H. Piwowarska-Bilska, A. Celler, B. Birkenfeld, Quanti-tative SPECT/CT reconstruction for 177Lu and 90Y targeted radionuclidetherapies, Physics in Medicine and Biology 57 (18) (2012) 5733–5747.doi:10.1088/0031-9155/57/18/5733.URL http://www.ncbi.nlm.nih.gov/pubmed/22948135[107] J.-M. Beauregard, M. S. Hofman, J. M. Pereira, P. Eu, R. J. Hicks,Quantitative (177)Lu SPECT (QSPECT) imaging using a commercially182Bibliographyavailable SPECT/CT system., Cancer imaging : the official publica-tion of the International Cancer Imaging Society 11 (2011) 56–66.doi:10.1102/1470-7330.2011.0012.URL http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3205754&tool=pmcentrez&rendertype=abstract[108] D. J. Kwekkeboom, W. W. De Herder, B. L. Kam, C. H. Van Eijck,M. Van Essen, P. P. Kooij, R. a. Feelders, M. O. Van Aken, E. P.Krenning, Treatment with the radiolabeled somatostatin analog [177Lu-DOTA0,Tyr3]octreotate: Toxicity, efficacy, and survival, Journal of Clin-ical Oncology 26 (13) (2008) 2124–2130. doi:10.1200/JCO.2007.15.2553.[109] S. Ezziddin, F. Khalaf, M. Vanezi, T. Haslerud, K. Mayer, A. Al Zreiqat,W. Willinek, H. J. Biersack, A. Sabet, Outcome of peptide receptor ra-dionuclide therapy with 177Lu-octreotate in advanced grade 1/2 pancre-atic neuroendocrine tumours, European Journal of Nuclear Medicine andMolecular Imaging (2014) 1–9doi:10.1007/s00259-013-2677-3.[110] A. Sabet, T. Haslerud, U. F. Pape, A. Sabet, H. Ahmadzadehfar, F. Grün-wald, S. Guhlke, H. J. Biersack, S. Ezziddin, Outcome and toxicity ofsalvage therapy with 177Lu-octreotate in patients with metastatic gas-troenteropancreatic neuroendocrine tumours, European Journal of NuclearMedicine and Molecular Imaging 41 (2) (2014) 205–210. doi:10.1007/s00259-013-2547-z.183Bibliography[111] A. D. Puranik, H. R. Kulkarni, A. Singh, R. P. Baum, Peptide recep-tor radionuclide therapy with 90Y/177Lu-labelled peptides for inopera-ble head and neck paragangliomas (glomus tumours), European Journalof Nuclear Medicine and Molecular Imaging 42 (8) (2015) 1223–1230.doi:10.1007/s00259-015-3029-2.URL http://link.springer.com/10.1007/s00259-015-3029-2[112] P. Radojewski, R. Dumont, N. Marincek, P. Brunner, H. R. Mäcke,J. Müller-Brand, M. Briel, M. a. Walter, Towards tailored radiopeptide ther-apy, European Journal of Nuclear Medicine and Molecular Imaging 42 (8)(2015) 1231–1237. doi:10.1007/s00259-015-3030-9.URL http://link.springer.com/10.1007/s00259-015-3030-9[113] K. Sjogreen Gleisner, G. Brolin, a. Sundlov, E. Mjekiqi, K. Ostlund,J. Tennvall, E. Larsson, Long-term retention of 177Lu/177mLu-Dotatatein patients investigated by gamma spectrometry and gamma cam-era imaging, Journal of Nuclear Medicine 56 (7) (2015) 976–985.doi:10.2967/jnumed.115.155390.URL http://jnm.snmjournals.org/cgi/doi/10.2967/jnumed.115.155390[114] L. Strigari, M. Konijnenberg, C. Chiesa, M. Bardies, Y. Du, K. S. Gleis-ner, M. Lassmann, G. Flux, The evidence base for the use of internaldosimetry in the clinical practice of molecular radiotherapy, European Jour-184Bibliographynal of Nuclear Medicine and Molecular Imaging (2014) 1976–1988doi:10.1007/s00259-014-2824-5.[115] Y. K. Dewaraja, S. J. Wilderman, M. Ljungberg, K. F. Koral, K. Za-sadny, M. S. Kaminiski, Accurate dosimetry in 131I radionuclide ther-apy using patient-specific, 3-dimensional methods for SPECT reconstruc-tion and absorbed dose calculation., Journal of nuclear medicine : of-ficial publication, Society of Nuclear Medicine 46 (5) (2005) 840–849.doi:46/5/840[pii].[116] M. Ljungberg, K. Sjögreen, X. Liu, E. Frey, Y. Dewaraja, S.-E. Strand,A 3-dimensional absorbed dose calculation method based on quantitativeSPECT for radionuclide therapy: evaluation for (131)I using monte carlosimulation., Journal of nuclear medicine : official publication, Society ofNuclear Medicine 43 (8) (2002) 1101–1109.[117] J. Zeintl, A. H. Vija, A. Yahil, J. Hornegger, T. Kuwert, Quantitative accu-racy of clinical 99mTc SPECT/CT using ordered-subset expectation max-imization with 3-dimensional resolution recovery, attenuation, and scattercorrection., Journal of nuclear medicine : official publication, Society ofNuclear Medicine 51 (6) (2010) 921–928. doi:10.2967/jnumed.109.071571.[118] S. Shcherbinin, a. Celler, T. Belhocine, R. Vanderwerf, a. Driedger, Ac-curacy of quantitative reconstructions in SPECT/CT imaging., Physics185Bibliographyin medicine and biology 53 (17) (2008) 4595–4604. doi:10.1088/0031-9155/53/17/009.[119] E. C. Frey, J. L. Humm, M. Ljungberg, Accuracy and Precision of Ra-dioactivity Quantification in Nuclear Medicine Images, Semin Nucl Med42 (3) (2013) 208–218. doi:10.1053/j.semnuclmed.2011.11.003.Accuracy.[120] N. Anizan, H. Wang, X. C. Zhou, R. F. Hobbs, R. L. Wahl, E. C. Frey,Factors affecting the stability and repeatability of gamma camera calibra-tion for quantitative imaging applications based on a retrospective reviewof clinical data, EJNMMI Research 4 (1) (2014) 1–11. doi:10.1186/s13550-014-0067-x.URL http://www.ejnmmires.com/content/4/1/67[121] N. Anizan, H. Wang, X. C. Zhou, R. L. Wahl, E. C. Frey, Factors affectingthe repeatability of gamma camera calibration for quantitative imagingapplications using a sealed source, Physics in Medicine and Biology 60 (3)(2015) 1325–1337. doi:10.1088/0031-9155/60/3/1325.URL http://stacks.iop.org/0031-9155/60/i=3/a=1325?key=crossref.cfafdf86ad217cad1c93605554fdd0cd[122] R. de Nijs, V. Lagerburg, T. L. Klausen, S. r. Holm, Improving quantitativedosimetry in 177Lu-DOTATATE SPECT by energy window-based scattercorrections., Nuclear medicine communications 35 (5) (2014) 522–33.186Bibliographydoi:10.1097/MNM.0000000000000079.URL http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3969156&tool=pmcentrez&rendertype=abstract[123] J. C. Sanders, T. Kuwert, J. Hornegger, P. Ritt, Quantitative SPECT/CTImaging of 177Lu with In Vivo Validation in Patients Undergoing PeptideReceptor Radionuclide Therapy, Molecular Imaging and Biologydoi:10.1007/s11307-014-0806-4.URL http://link.springer.com/10.1007/s11307-014-0806-4[124] E. Vandervoort, A. Celler, R. Harrop, Implementation of an iterative scat-ter correction, the influence of attenuation map quality and their effect onabsolute quantitation in SPECT., Physics in medicine and biology 52 (5)(2007) 1527–1545. doi:10.1088/0031-9155/52/5/020.URL http://www.ncbi.nlm.nih.gov/pubmed/17301469[125] S. Shcherbinin, J. Grimes, a. Bator, J. B. Cwikla, a. Celler, Three-dimensional personalized dosimetry for (188)Re liver selective internal ra-diation therapy based on quantitative post-treatment SPECT studies., Phys.Med. Biol. 59 (1) (2013) 119–134. doi:10.1088/0031-9155/59/1/119.URL http://www.ncbi.nlm.nih.gov/pubmed/24334821[126] Y. E. Erdi, B. W. Wessels, M. H. Loew, a. K. Erdi, Threshold estimation insingle photon emission computed tomography and planar imaging for clin-ical radioimmunotherapy., Cancer research 55 (23 Suppl) (1995) 5823s–5826s.187Bibliography[127] J. Grimes, A. Celler, S. Shcherbinin, H. Piwowarska-Bilska, B. Birken-feld, The accuracy and reproducibility of SPECT target volumes and activ-ities estimated using an iterative adaptive thresholding technique., Nuclearmedicine communications 33 (12) (2012) 1254–1266. doi:10.1097/MNM.0b013e3283598395.URL http://www.ncbi.nlm.nih.gov/pubmed/23010981[128] J. E. Arnold, a. S. Johnston, S. M. Pinsky, The influence of true countingrate and the photopeak fraction of detected events on Anger camera dead-time., Journal of nuclear medicine 15 (6) (1974) 412–416.[129] J. Sorenson, Deadtime characteristics of Anger cameras., Journal of nuclearmedicine: official publication, . . . 16 (4) (1975) 284–288.URL http://europepmc.org/abstract/MED/1113185[130] C. Chiesa, A. Negri, C. Albertini, R. Azzeroni, E. Setti, L. Mainardi, G. Al-iberti, E. Seregni, E. Bombardier, A practical dead time correction methodin planar activity ..., The Quarterly Journal of Nuclear Medicine and Molec-ular Imaging 53 (2009) 658–670.[131] G. Delpon, L. Ferrer, a. Lisbona, M. Bardiès, Correction of count losses dueto deadtime on a DST-XLi (SmVi-GE) camera during dosimetric studies inpatients injected with iodine-131., Physics in medicine and biology 47 (7)(2002) N79–N90. doi:10.1088/0031-9155/47/7/402.[132] D. Guirado, J. C. Ramírez, J. M. De la Vega, M. Vilches, a. M. Lallena,188BibliographyQuality control for system count rate performance with scatter in gammacameras, Physica Medica 28 (3) (2012) 254–261. doi:10.1016/j.ejmp.2011.05.002.[133] M. Silosky, V. Johnson, C. Beasley, S. C. Kappadath, Characterization ofthe count rate performance of modern gamma cameras., Medical physics40 (3) (2013) 032502. doi:10.1118/1.4792297.URL http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3829891&tool=pmcentrez&rendertype=abstract[134] A. Celler, H. Piwowarska-Bilska, S. Shcherbinin, C. Uribe, R. Mikolajczak,B. Birkenfeld, Evaluation of dead-time corrections for post-radionuclide-therapy (177)Lu quantitative imaging with low-energy high-resolution col-limators., Nuclear medicine communications 35 (1) (2014) 73–87. doi:10.1097/MNM.0000000000000011.URL http://www.ncbi.nlm.nih.gov/pubmed/24131941[135] R. Adams, G. J. Hine, C. D. Zimmerman, Deadtime Measurements in Scin-tillation Cameras under Scatter Conditions Simulating Quantitative NuclearCardiography, The Journal of Nuclear Medicine 19 (1978) 538–544.[136] K. R. Zasadny, K. F. Koral, F. M. Swailem, Dead time of an anger camerain dual-energy-window-acquisition mode., Medical physics 20 (4) (2005)1115–1120. doi:10.1118/1.597008.[137] V. Moiseenko, UBC PHYS-536: Advanced Radiation Biophysics (2010).189Bibliography[138] C. J. Hornby, Tumour Control and Normal Tissue Complication Proba-bilities : Can they be correlated with the measured clinical outcomes ofprostate cancer radiotherapy ?, Ph.D. thesis (2005).URL https://researchbank.rmit.edu.au/eserv/rmit:6695/Hornby.pdf[139] J.-P. Pouget, C. Lozza, E. Deshayes, V. Boudousq, I. Navarro-Teulon,Introduction to radiobiology of targeted radionuclide therapy., Front. Med.2 (March) (2015) 12. doi:10.3389/fmed.2015.00012.URL http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=4362338{&}tool=pmcentrez{&}rendertype=abstract[140] D. J. Brenner, L. R. Hlatky, P. J. Hahnfeldt, Y. Huang, R. K. Sachs, Thelinear-quadratic model and most other common radiobiological models re-sult in similar predictions of time-dose relationships., Radiation research150 (1) (1998) 83–91. arXiv:NIHMS150003, doi:10.2307/3579648.[141] R. Dale, Use of the linear-quadratic radiobiological model for quantifyingkidney response in targeted radiotherapy., Cancer biotherapy & radiophar-maceuticals 19 (3) (2004) 363–70. doi:10.1089/1084978041425070.URL http://www.ncbi.nlm.nih.gov/pubmed/15285884[142] L. G. Bouchet, W. E. Bolch, H. P. Blanco, B. W. Wessels, J. a. Siegel, D. a.Rajon, I. Clairand, G. Sgouros, MIRD Pamphlet No 19: absorbed fractionsand radionuclide S values for six age-dependent multiregion models of the190Bibliographykidney., Journal of nuclear medicine : official publication, Society of Nu-clear Medicine 44 (7) (2003) 1113–1147.[143] M. G. Stabin, R. B. Sparks, E. Crowe, OLINDA/EXM: the second-generation personal computer software for internal dose assessment in nu-clear medicine., Journal of nuclear medicine : official publication, Societyof Nuclear Medicine 46 (6) (2005) 1023–7.URL http://www.ncbi.nlm.nih.gov/pubmed/15937315[144] M. G. Stabin, X. G. Xu, M. a. Emmons, W. P. Segars, C. Shi, M. J. Fernald,RADAR Reference Adult, Pediatric, and Pregnant Female Phantom Seriesfor Internal and External Dosimetry, Journal of Nuclear Medicine 53 (11)(2012) 1807–1813. doi:10.2967/jnumed.112.106138.[145] W. E. Bolch, L. G. Bouchet, J. S. Robertson, B. W. Wessels, J. a. Siegel,R. W. Howell, a. K. Erdi, B. Aydogan, S. Costes, E. E. Watson, a. B. Brill,N. D. Charkes, D. R. Fisher, M. T. Hays, S. R. Thomas, MIRD pamphletNo. 17: the dosimetry of nonuniform activity distributions–radionuclideS values at the voxel level. Medical Internal Radiation Dose Commit-tee., Journal of nuclear medicine : official publication, Society of NuclearMedicine 40 (1) (1999) 11S–36S.[146] N. Lanconelli, M. Pacilio, S. L. Meo, F. Botta, a. D. Dia, L. a. T. Aroche,M. a. C. Pérez, M. Cremonesi, A free database of radionuclide voxel Svalues for the dosimetry of nonuniform activity distributions, Physics in191BibliographyMedicine and Biology 57 (2) (2012) 517–533. doi:10.1088/0031-9155/57/2/517.[147] R. N. J. Graham, R. W. Perriss, a. F. Scarsbrook, DICOM demystified: areview of digital file formats and their use in radiological practice., Clinicalradiology 60 (11) (2005) 1133–40. doi:10.1016/j.crad.2005.07.003.URL http://www.ncbi.nlm.nih.gov/pubmed/16223609[148] S. Blinder, A. Celler, R. Wells, D. Thomson, R. Harrop, Experimen-tal verification of 3D detector response compensation using the OSEMreconstruction method, 2001 IEEE Nuclear Science Symposium Confer-ence Record (Cat. No.01CH37310) 4 (2001) 2174–2178. doi:10.1109/NSSMIC.2001.1009254.URL http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=1009254[149] B. F. Hutton, I. Buvat, F. J. Beekman, Review and current status of SPECTscatter correction., Physics in medicine and biology 56 (14) (2011) R85–112. doi:10.1088/0031-9155/56/14/R01.URL http://www.ncbi.nlm.nih.gov/pubmed/21701055[150] R. J. Jaszczak, K. L. Greer, C. E. Floyd, C. C. Harris, R. E. Coleman,J. Jaszczak, E. Floyd, C. Craig, Improved SPECT Quantification UsingCompensation for Scattered Photons ImprovedSPECTQuantification Us-ingCompensation for ScatteredPhotons, Journal of Nuclear Medicine 25(1984) 893–900.192Bibliography[151] S. Shcherbinin, a. Celler, T. Belhocine, R. Vanderwerf, a. Driedger, Ac-curacy of quantitative reconstructions in SPECT/CT imaging., Physicsin medicine and biology 53 (17) (2008) 4595–4604. doi:10.1088/0031-9155/53/17/009.URL http://www.ncbi.nlm.nih.gov/pubmed/18678930193

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:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0223889/manifest

Comment

Related Items