Single photon emission computed tomography (SPECT) is used in a variety of medical procedures, including myocardial perfusion, bone metabolism, and thyroid function studies. In SPECT, the emissions of a radionuclide within a patient are counted at a gamma camera to form a 2-dimensional projection of the 3-dimensional radionuclide distribution within the patient. This unknown 3-dimensional source distribution can be reconstructed from many 2-dimensional projections obtained at different angles around the patient. This reconstruction can be improved by properly modeling the physics in the patient, i.e., particle absorption and scattering. Currently, such modeling is done using statistical Monte Carlo methods, but deterministic codes have the potential to offer fast computation speeds while fully modeling particle interactions within the patient. Deterministic codes are not susceptible to statistical uncertainty, but have been over-looked for applications to nuclear medicine, most likely due to their own limitations, including discretization and large memory requirements. A novel deterministic reconstruction methodology for SPECT (DRS) has been developed to apply the advantages of deterministic algorithms to SPECT iterative image reconstruction. Using a maximum likelihood expectation maximization (ML-EM) algorithm, a deterministic code can fully model particle transport in the patient in the forward projection step, without the need of a large system matrix. The TITAN deterministic transport code has a SPECT formulation that allows for fast simulation of SPECT projection images and has been benchmarked through comparison with results from the SIMIND and MCNP5 Monte Carlo codes in this dissertation. The TITAN SPECT formulation has been improved through a modified collimator representation and full parallelization. The DRS methodology has been implemented in the TITAN code to create TITAN with Image Reconstruction (TITAN-IR). The TITAN-IR code has been used to successfully reconstruct the source distribution from SPECT data for the Jaszczak and NCAT phantoms. Extensive studies have been conducted to examine the sensitivity of TITAN-IR image quality to deterministic parameter selection as well as collimator blur and noise in the projection data being reconstructed. The TITAN-IR reconstruction has also been compared with other reconstruction algorithms. This novel image reconstruction methodology has been shown to reconstruct images in short computation times, demonstrating its potential in a clinical setting with further development. / Ph. D.

With the recent interest in single fraction Stereotactic Body Radiation Therapy and the emerging prominence of the Rapid Arc radiotherapy technique capable of delivering a fast and accurate treatment, the in-field primary dose and out-of-field dose assessments are becoming increasingly important. Currently, full physics Monte Carlo calculations for dose calculations have been regarded as a ‘gold standard’ for dose assessments of the target and OAR (organ at risk). However, these Monte Carlo calculations require very long computation times. The current treatment planning methods provide shorter calculation times, but issues such as heterogeneities and model-based parameter calculations cause challenges and affect dose calculation accuracy. This thesis describes a new and fast dose estimation method leveraging parallel computing called EDK-SN, “Electron Dose Kernel-Discrete Ordinates”. This new method uses hybrid electron dose kernels driven by linear Boltzmann (discrete ordinates) photon transport method to carry out dose calculations. The method has proven effective for fast and accurate computations of out-of-field whole body dose calculations benchmarked to Monte Carlo with isotropic monoenergetic photon sources. This thesis accomplishes adaptation of clinical Varian phase space data for use with general Monte Carlo codes including MCNP, and mapping accurate phase space data into the application of optimized EDK-SN dose calculation method with a 15-year-old patient phantom. The EDK-SN method with improved source term modeling is demonstrated to fall within accuracy of the measured golden beam data for a clinical water phantom.

Depuis deux décennies, les nombreux développements des techniques de radiothérapie par modulation d’intensité (RCMI) ont permis de mieux conformer la dose au volume cible et ainsi, d’augmenter les taux de réussite des traitements des cancers. Ces techniques ont souvent l’avantage de réduire la dose aux organes à risque proches de la zone traitée, mais elles ont l’inconvénient d’apporter un niveau de dose périphérique plus important que les techniques basiques sans modulation d’intensité. Dans ce contexte, l’augmentation du taux de survie des patients qui en résulte, accroît également la probabilité de manifestation d’effets iatrogènes dus aux doses périphériques (tels que les cancers secondaires). Aujourd’hui, la dose périphérique n’est pas considérée lors de la planification du traitement et il n’existe aucun outil numérique fiable pour sa prédiction. Il devient cependant indispensable de prendre en compte le dépôt de dose périphérique lors de la planification du traitement, notamment dans les cas pédiatriques. Cette étude doctorale a permis la réalisation de plusieurs étapes du développement d’un outil numérique, précis et rapide, de prédiction de la dose hors-champ fondé sur le code Monte Carlo PENELOPE. Dans cet objectif, nous avons démontré la capacité du code PENELOPE à estimer la dose périphérique en comparant ses résultats avec des mesures de référence réalisées à partir de deux configurations expérimentales (métrologique et pré-clinique). Ces travaux expérimentaux ont notamment permis la mise en place d’un protocole d’utilisation des dosimètres OSL pour la mesure des faibles doses. En parallèle, nous avons pu mettre en évidence la convergence lente et rédhibitoire du calcul en vue d’une utilisation clinique. Par conséquent nous avons réalisé un travail d’accélération du code en implémentant une nouvelle technique de réduction de variance appelée transport pseudo-déterministe spécifiquement dédiée à l’amélioration de la convergence dans des zones lointaines du faisceau principal. Ces travaux ont permis d’améliorer l’efficacité des estimations dans les deux configurations de validation définies (gain d’un facteur 20) pour atteindre des temps de calcul raisonnables pour une application clinique. Des travaux d’optimisation du code restent à entreprendre de façon à améliorer encore la convergence de l’outil pour ensuite en envisager une utilisation clinique. / Over the last two decades, many technical developments have been achieved on intensity modulated radiotherapy (IMRT) and allow a better conformation of the dose to the tumor and consequently increase the success of cancer treatments. These techniques often reduce the dose to organs at risk close to the target volume; nevertheless they increase peripheral dose levels. In this situation, the rising of the survival rate also increases the probability of secondary effects expression caused by peripheral dose deposition (second cancers for instance). Nowadays, the peripheral dose is not taken into account during the treatment planification and no reliable prediction tool exists. However it becomes crucial to consider the peripheral dose during the planification, especially for pediatric cases. Many steps of the development of an accurate and fast Monte Carlo out-of-field dose prediction tool based on the PENELOPE code have been achieved during this PhD work. To this end, we demonstrated the ability of the PENELOPE code to estimate the peripheral dose by comparing its results with reference measurements performed on two experimental configurations (metrological and pre-clinical). During this experimental work, we defined a protocol for low doses measurement with OSL dosimeters. In parallel, we highlighted the slow convergence of the code for clinical use. Consequently, we accelerated the code by implementing a new variance reduction technique called pseudo-deterministic transport which is specifically with the objective of improving calculations in areas far away from the beam. This step improved the efficiency of the peripheral doses estimation in both validation configurations (by a factor of 20) in order to reach reasonable computing times for clinical application. Optimization works must be realized in order improve the convergence of our tool and consider a final clinical use.

The accurate and efficient simulation of coupled neutron-photon problems is necessary for several important radiation detection applications. Examples include the detection of nuclear threats concealed in cargo containers and prompt gamma neutron activation analysis for nondestructive determination of elemental composition of unknown samples. In these applications, high-resolution gamma-ray spectrometers are used to preserve as much information as possible about the emitted photon flux, which consists of both continuum and characteristic gamma rays with discrete energies. Monte Carlo transport is the most commonly used modeling tool for this type of problem, but computational times for many problems can be prohibitive. This work explored the use of coupled Monte Carlo-deterministic methods for the simulation of neutron-induced photons for high-resolution gamma-ray spectroscopy applications. A method was developed for the implementation of coupled neutron-photon problems into RAdiation Detection Scenario Analysis Toolbox (RADSAT), a computer code that couples the complementary strengths of discrete-ordinate and Monte Carlo approaches to obtain high-resolution detector responses. Central to this work was the development of a method for generating multi-group neutron-photon cross-sections in a way that separates the discrete and continuum photon emissions so that the key signatures in neutron activation analysis (i.e., the characteristic line energies) are preserved. The mechanics of the cross-section preparation method are described and contrasted with standard neutron-gamma cross-section sets. These custom cross-sections were then applied to several benchmark problems using the method developed in this work. Multi-group results for neutron and photon flux are compared to MCNP results. Finally, calculated responses of high-resolution spectrometers were compared. The added computational efficiency of the coupled Monte Carlo-deterministic method and the positive agreement achieved in the code-to-code verification make the integration of the coupled neutron-photon method into RADSAT a promising endeavor.