Cancer is a group of diseases that are characterized by abnormal cell growth and is considered a leading cause of death globally. There are a number of different cancer treatment modalities, one of which is radiation therapy. In radiation therapy treatment planning, it is important to make sure that enough radiation is delivered to the tumor and that healthy organs are spared, while also making sure to account for uncertainties such as misalignment of the patient during treatment. To reduce the workload on clinics, data-driven automated treatment planning can be used to generate treatment plans for new patients based on previously delivered plans. In this thesis, we propose a novel method for robust automated treatment planning where a deep learning model is trained to deform a dose in accordance with a set of potential scenarios that account for the different uncertainties while maintaining certain statistical properties of the input dose. The predicted scenario doses are then used in a robust optimization problem with the goal of finding a treatment plan that is robust to these uncertainties. The results show that the proposed method for deforming doses yields realistic doses of high quality and that the proposed pipeline can potentially generate doses that conform better to the target than the current state of the art but at the cost of dose homogeneity. / Cancer är ett samlingsnamn för sjukdomar som karaktäriseras av onormal celltillväxt och betraktas som en ledande dödsorsak globalt. Det finns olika typer av cancerbehandling, varav en är strålterapi. Inom strålterapiplanering är det viktigt att säkerställa att tillräckligt med strålning ges till tumören, att friska organ skonas, och att osäkerheter som felplacering av patienten under behandlingen räknas med. För att minska arbetsbelastningen på kliniker används data-driven automatisk strålterapiplanering för att generera behandlingsplaner till nya patienter baserat på tidigare levererade behandlingar. I denna uppsats föreslår vi en ny metod för robust automatisk strålterapiplanering där en djupinlärningsmodell tränas till att deformera en dos i enlighet med en mängd potentiella scenarion som motsvarar de olika osäkerheterna medan vissa statistiska egenskaper bibehålls från originaldosen. De predicerade scenariodoserna används sedan i ett robust optimeringsproblem där målet är att hitta en behandlingsplan som är robust mot dessa osäkerheter. Resultaten visar att den föreslagna metoden för dosdeformation ger realistiska doser av hög kvalitet, vilket i sin tur kan leda till robusta doser med högre doskonformitet än tidigare metoder men på bekostnad av doshomogenitet.

Aims Use of 3D printed models in Endodontics has been gaining popularity since the technology to create them became more affordable. Currently, there are no studies that evaluate the influence of 3D models on endodontic surgical treatment planning and on operator confidence. Therefore, aims of this study were to: (i) Determine whether the availability of a 3D printed analogue can influence treatment-planning and operator confidence; and, (ii) Assess which factors of operator confidence are influenced, if any. Materials and Methods Endodontists were asked to analyze a pre-selected CBCT scan of an endodontic surgical case and to answer a questionnaire that determined their surgical approach for that case. After 30 days, the same participants were asked to analyze again the same CBCT scan. This time however, a 3D printed model of the scan was made available to the participants and to perform a mock osteotomy on the model. The participants were then asked to respond to the same questionnaire that they responded to 30 days prior to determine if there would be any changes to their treatment plan. A new set of questions were added to the survey to evaluate the influence of the 3D printed model on participants’ confidence in performing endodontic surgery. The responses were statistically analyzed using Chi square test followed by either logistic or ordered regression analysis while adjusting for experience of participant. Adjustment for multiple comparison analysis was done using Bonferroni correction. Statistical significance was set at £0.005. Results Availability of the 3D printed model and the CBCT scan together resulted in statistically significant differences in the participants’ responses to their ability to clearly detect bone landmarks, accurately predict the location of osteotomy, and in determining the following: size of osteotomy, angle of instrumentation, involvement of critical structures in flap reflection and involvement of vital structures during curettage. In addition, the participants’ confidence in performing surgery was significantly higher versus having CBCT scans alone. There were no statistically significant changes with decisions on flap design and extent, visualizing critical structures, lesion size, injury to vital structures during osteotomy, the length of root that could be resected and the number of roots involved. Conclusions The availability of 3D printed models did not alter the participants’ surgical approach, but it significantly improved their confidence for endodontic microsurgery, which can be attributed to better visualization of anatomical structures.

Irreversible electroporation is a minimally invasive technique for the non-thermal destruction of cells in a targeted volume of tissue, using brief electric pulses, (~100 µs long) delivered through electrodes placed into or around the targeted region. These electric pulses destabilize the integrity of the cell membrane, resulting in the creation of nanoscale defects that increase a cell’s permeability to exchange with its environment. When the energy of the pulses is high enough, the cell cannot recover from these effects and dies in a non-thermal manner that does not damage neighboring structures, including the extracellular matrix. IRE has been shown to spare the major vasculature, myelin sheaths, and other supporting tissues, permitting its use in proximity to these vital structures. This technique has been proposed to be harnessed as an advantageous non-thermal focal ablation technique for diseased tissues, including tumors. IRE electric pulses may be delivered through small (ø ≈ 1 mm) needle electrodes, making treatments minimally invasive and easy to apply. There is sub-millimeter demarcation between treated and unaffected cells, which may be correlated with the electric field to which the tissue is exposed, enabling numerical predictions to facilitate treatment planning. Immediate changes in the cellular and tissue structure allow real-time monitoring of affected volumes with imaging techniques such as computed tomography, magnetic resonance imaging, electrical impedance tomography, or ultrasound. The ability to kill tumor cells has been shown to be independent of a functioning immune system, though an immune response seems to be promoted by the ablation. Treatments are unaltered by blood flow and the electric pulses may be administered quickly (~ 5 min). Recently, safety and case studies using IRE for tumor therapy in animal and human patients have shown promising results. Apart from these new studies, previous work with IRE has involved studies in healthy tissues and small cutaneous experimental tumors. As a result, there remain significant differences that must be considered when translating this ablation technique towards a successful and reliable therapeutic option for patients. The dissertation work presented here is designed to develop irreversible electroporation into a robust, clinically viable treatment modality for targeted regions of diseased tissue, with an emphasis on tumors. This includes examining and creating proving the efficacy for IRE therapy when presented with the many complexities that present themselves in real-world clinical patient therapies, including heterogeneous environments, large and irregular tumor geometries, and dynamic tissue properties resulting from treatment. The impact of these factors were theoretically tested using preliminary in vitro work and numerical modeling to determine the feasibility of IRE therapy in heterogeneous systems. The feasibility of use was validated in vivo with the successful treatment of human mammary carcinomas orthotopically implanted in the mammary fat pad of mice using a simple, single needle electrode design easily translatable to clinical environments. Following preliminary theoretical and experimental work, this dissertation considers the most effective and accurate treatment planning strategies for developing optimal therapeutic outcomes. It also experimentally characterizes the dynamic changes in tissue properties that result from the effects of IRE therapy using ex vivo porcine renal cortical tissue and incorporates these into a revised treatment planning model. The ability to use the developments from this earlier work is empirically tested in the treatment of a large sarcoma in a canine patient that was surgically unresectable due to its proximity to critical arteries and the sciatic nerve. The tumor was a large and irregular shape, located in a heterogeneous environment. Treatment planning was performed and the therapy carried out, ultimately resulting in the patient being in complete remission for 14 months at the time of composing this work. The work presented in this dissertation finishes by examining potential supplements to enhance IRE therapy, including the presence of an inherent tumor-specific patient immune response and the addition of adjuvant therapeutic modalities. / Ph. D.

Le Cyberknife (Accuray, Sunnyvale, CA) est un appareil de radiochirurgie stéréotaxique sans cadre. Il a été développé pour administrer de fortes doses dans des volumes restreints. Aussi, pour obtenir une conformation optimale de traitement, des champs circulaires de petites dimensions sont utilisés (\phi = 0,5 à 6 cm). L'étude dosimétrique de ces petits champs doit être menée selon de nouveaux standards puisque ceux-ci échappent aux définitions du TG-51. L'objectif de ce projet est d'implanter une plateforme de calcul de dose de type Monte Carlo pour le CyberKnife en clinique. Il s'articule autour de deux réalisations principales. Tout d'abord, une caractérisation dosimétrique du modèle Monte Carlo de l'accélérateur linéaire du CyberKnife a été menée à travers des simulations Monte Carlo générées par le moteur de EGSnrc. Cette étude est basée sur la caractérisation de la réponse d'un détecteur à un champ de type CK à partir de simulations EGS_chamber. Cette approche permet de prendre en compte l'impact du détecteur sur les mesures expérimentales. Cet aspect est d'autant plus important que le modèle Monte Carlo de l'accélérateur est validé à partir de mesures expérimentales. Les résultats obtenus montrent une bonne concordance, <1% ou 1 mm, entre les mesures expérimentales et les données de simulations pour les grands champs. Pour les champs de diamètre < 12,5 mm, le modèle est moins exact et une correction est appliquée pour atteindre une différence de <1% ou 1 mm. Deuxièmement, ce modèle validé du CK a été implanté dans un cadre de calcul Monte Carlo complet. Une plateforme de calcul dédiée aux calculs Monte Carlo, WebTPS, a été adaptée aux calculs de dose CK. Cette plateforme reçoit les données relatives au plan de traitement et lance des calcul EGSnrc sur un superordinateur. Cette approche tend à réduire les approximations lors de l'évaluation dosimétrique de plans de traitements cliniques. Une incertitude inférieure à 1% peut être atteinte en deux heures de calcul. Ce projet a donc pour objectif de développer une référence clinique pour le calcul de dose dans le cadre de la radiochirurgie stéréotaxique. L'outil WebTPS pourrait être particulièrement utile en clinique, l'algorithme de calcul de dose du CK étant limité dans plusieurs situations de traitement. / Purpose: The scope of this study is to implement a clinical Monte Carlo dose calculation system based on the EGSnrc engine. This web-based tool will be mostly used to evaluate clinical treatment plans in highly heterogeneous phantoms. Methods: The Monte Carlo calculation tool is based on the DOSXYZnrc user code. The platform automatically converts CyberKnife clinical plan to the user code input files. Phantoms can be created from HU to ED curves or by manually assigning material using medical contours. Parallel computation is made on a Compute Canada high-performance cluster to reduce simulation time. A Monte Carlo CyberKnife model is built on BEAMnrc user code using the manufacturer specifications. Simulated and experimental data is compared to estimate the electron beam parameters. The beam energy estimation is based on percent depth dose (PDD) comparison while the full width at half max (FWHM) is validated by output factor (OF) and off-axis ratio (OAR). An EGS_chamber model of the PTW60012 diode is used in the OF calculation. A set of phase-spaces is generated from the optimal model and for each collimator to calculate dose contribution from each incident beam. Results: The linac model optimisation yielded a 0.5% PDD agreement between experimental and simulation data, and a 0.5% or 1 mm for OAR. DOSxyz simulation of full treatment plan, based on the preliminary CyberKnife model, were achieved. Total Monte Carlo dose calculation have been achieved for heterogeneous phantoms. Uncertainty under 1% can be achieved for less than 2 hour of computing time. However, computing time estimation is nontrivial due to its dependence on cluster availability. Conclusion: This work aims to develop a suitable tool for reference plan dose calculation. This web-based tool would be used in several clinical and research applications where the CyberKnife embedded ray-tracing algorithm would show significant limitations. Because it is destined to a clinical use, the whole dose calculation system will be rigorously validated. / Le travail de modélisation a été réalisé à travers EGSnrc, un logiciel développé par le Conseil National de Recherche Canada.

Le travail de modélisation a été réalisé à travers EGSnrc, un logiciel développé par le Conseil National de Recherche Canada. / Le Cyberknife (Accuray, Sunnyvale, CA) est un appareil de radiochirurgie stéréotaxique sans cadre. Il a été développé pour administrer de fortes doses dans des volumes restreints. Aussi, pour obtenir une conformation optimale de traitement, des champs circulaires de petites dimensions sont utilisés (\phi = 0,5 à 6 cm). L'étude dosimétrique de ces petits champs doit être menée selon de nouveaux standards puisque ceux-ci échappent aux définitions du TG-51. L'objectif de ce projet est d'implanter une plateforme de calcul de dose de type Monte Carlo pour le CyberKnife en clinique. Il s'articule autour de deux réalisations principales. Tout d'abord, une caractérisation dosimétrique du modèle Monte Carlo de l'accélérateur linéaire du CyberKnife a été menée à travers des simulations Monte Carlo générées par le moteur de EGSnrc. Cette étude est basée sur la caractérisation de la réponse d'un détecteur à un champ de type CK à partir de simulations EGS_chamber. Cette approche permet de prendre en compte l'impact du détecteur sur les mesures expérimentales. Cet aspect est d'autant plus important que le modèle Monte Carlo de l'accélérateur est validé à partir de mesures expérimentales. Les résultats obtenus montrent une bonne concordance, <1% ou 1 mm, entre les mesures expérimentales et les données de simulations pour les grands champs. Pour les champs de diamètre < 12,5 mm, le modèle est moins exact et une correction est appliquée pour atteindre une différence de <1% ou 1 mm. Deuxièmement, ce modèle validé du CK a été implanté dans un cadre de calcul Monte Carlo complet. Une plateforme de calcul dédiée aux calculs Monte Carlo, WebTPS, a été adaptée aux calculs de dose CK. Cette plateforme reçoit les données relatives au plan de traitement et lance des calcul EGSnrc sur un superordinateur. Cette approche tend à réduire les approximations lors de l'évaluation dosimétrique de plans de traitements cliniques. Une incertitude inférieure à 1% peut être atteinte en deux heures de calcul. Ce projet a donc pour objectif de développer une référence clinique pour le calcul de dose dans le cadre de la radiochirurgie stéréotaxique. L'outil WebTPS pourrait être particulièrement utile en clinique, l'algorithme de calcul de dose du CK étant limité dans plusieurs situations de traitement. / Purpose: The scope of this study is to implement a clinical Monte Carlo dose calculation system based on the EGSnrc engine. This web-based tool will be mostly used to evaluate clinical treatment plans in highly heterogeneous phantoms. Methods: The Monte Carlo calculation tool is based on the DOSXYZnrc user code. The platform automatically converts CyberKnife clinical plan to the user code input files. Phantoms can be created from HU to ED curves or by manually assigning material using medical contours. Parallel computation is made on a Compute Canada high-performance cluster to reduce simulation time. A Monte Carlo CyberKnife model is built on BEAMnrc user code using the manufacturer specifications. Simulated and experimental data is compared to estimate the electron beam parameters. The beam energy estimation is based on percent depth dose (PDD) comparison while the full width at half max (FWHM) is validated by output factor (OF) and off-axis ratio (OAR). An EGS_chamber model of the PTW60012 diode is used in the OF calculation. A set of phase-spaces is generated from the optimal model and for each collimator to calculate dose contribution from each incident beam. Results: The linac model optimisation yielded a 0.5% PDD agreement between experimental and simulation data, and a 0.5% or 1 mm for OAR. DOSxyz simulation of full treatment plan, based on the preliminary CyberKnife model, were achieved. Total Monte Carlo dose calculation have been achieved for heterogeneous phantoms. Uncertainty under 1% can be achieved for less than 2 hour of computing time. However, computing time estimation is nontrivial due to its dependence on cluster availability. Conclusion: This work aims to develop a suitable tool for reference plan dose calculation. This web-based tool would be used in several clinical and research applications where the CyberKnife embedded ray-tracing algorithm would show significant limitations. Because it is destined to a clinical use, the whole dose calculation system will be rigorously validated.

Les néoplasies pulmonaires demeurent la première cause de décès par cancer au Québec représentant près de 6000 décès par année. Au cours des dernières années, la radiothérapie stéréotaxique d’ablation (SABR) s’est imposée comme un traitement alternatif à la résection anatomique pour les patients inopérables atteints d’un cancer pulmonaire non à petites cellules de stade précoce. Il s’agit d’une modalité de traitement qui permet d’administrer des doses élevées, typiquement 30-60 Gy en 1-8 fractions, dans le but de cibler précisément le volume de traitement tout en épargnant les tissus sains. Le Centre Hospitalier de l’Université de Montréal s’est muni en 2009 d’un appareil de SABR de fine pointe, le CyberKnife™ (CK), un accélérateur linéaire produisant un faisceau de photons de 6 MV dirigé par un bras robotisé, permettant d’administrer des traitements non-coplanaires avec une précision infra-millimétrique. Ce mémoire est dédié à la caractérisation de certains enjeux cliniques et physiques associés au traitement par CK. Il s’articule autour de deux articles scientifiques revus par les pairs. D’une part, une étude prospective clinique présentant les avantages de la SABR pulmonaire, une technique qui offre un excellent contrôle tumoral à long terme et aide au maintien de la qualité de vie et de la fonction pulmonaire. D’autre part, une étude de physique médicale illustrant les limites de l’acquisition d’images tomodensitométriques en auto-rétention respiratoire lors de la planification de traitement par CK. / Lung neoplasia remains the leading cause of cancer death accounting for nearly 6,000 deaths per year in Quebec. In recent years, stereotactic ablative radiotherapy (SABR) has emerged as an alternative treatment to anatomical resection for inoperable patients suffering from early stage non-small cell lung cancer. This technique can deliver highly focused doses such as 30-60 Gy in 1-8 fractions in order to target precisely the treatment volume while sparing healthy tissue. In 2009, the Centre Hospitalier de l’Université de Montréal acquired a robotic SABR apparatus, the CyberKnife™ (CK), a linear accelerator mounted on a moving arm producing non-coplanar photon beams of 6 MV with millimetric precision. This thesis presents two scientific peer reviewed articles adressing some clinical and physical challenges with CK. On one hand, a clinical prospective study reporting the advantages of lung SABR, a technic that offers excellent long-term tumor control and helps maintain the quality of life and lung function. On the other hand, a medical physics study exposing the limits of computed tomography scan acquisition in breath-holding for CK treatment planning.

L'étalonnage de l’image tomodensitométrique X en pouvoirs d’arrêt relatif est source d'incertitudes pour la planification du traitement en protontherapie. L’imagerie proton permettrait d’obtenir directement les pouvoirs d’arrêt ou les épaisseurs équivalent-eau tout en maîtrisant les incertitudes sur la planification du traitement. Ce travail vise à caractériser et optimiser les performances du système de tomographie proton proposé dans le cadre d’une nouvelle approche faisceau à faisceau, composé d’un trajectographe et d’un range meter. La position et la largeur du faisceau obtenues avec le trajectographe ainsi que la modélisation matricielle de la réponse du range meter par simulation Monte Carlo combinée à la déconvolution de la courbe de Bragg résiduelle ont permis de reconstruire l’épaisseur équivalent-eau traversée pour chaque faisceau. L’évaluation de la qualité des images a permis de montrer que la méthode de déconvolution permettait d’obtenir des images dépourvues d’artefacts et d’estimer le parcours du proton avec une précision de l’ordre de 0,7%. Le travail présenté dans cette thèse démontre la faisabilité d’un tel système d’imagerie. / Calibration of computed tomography image in relative stopping power is a source of uncertainties for the proton therapy treatment planning. Proton imaging could directly obtain stopping powers or equivalent water thicknesses and control uncertainties in the treatment planning. In the context of a new pencil beam approach, this work aims to characterize and optimize the performances of a proton tomography system consisting of a tracker and a range meter. Beam position and width obtained with the tracker and the range meter response matrix modelling by Monte Carlo simulation combined with the unfolding method of the residual Bragg curve enable to reconstruct the water equivalent thickness for each beam. The evaluation of the reconstructed images quality shows that images are artefact free and the proton range is estimated with 0.7% of accurancy by using the unfolding method. This thesis demonstrates the feasibility of such an imaging system.

No description available.

We study inverse treatment planning approaches for stereotactic radiosurgery using Leksell Gamma Knife Perfexion (PFX, Elekta, Stockholm, Sweden) to treat brain cancer and tumour patients. PFX is a dedicated head-and-neck radiation delivery device that is commonly used in clinics. In a PFX treatment, the patient lies on a couch and the radiation beams are emitted from eight banks of radioactive sources around the patient's head that are focused at a single spot, called an isocentre. The radiation delivery in PFX follows a step-and-shoot manner, i.e., the couch is stationary while the radiation is delivered at an isocentre location, and only moves when no beam is being emitted. To find a set of well-positioned isocentres in tumour volumes, we explore fast geometry-based algorithms, including skeletonization and hybrid grassfire and sphere-packing approaches. For the selected set of isocentres, the optimal beam durations to deliver a high prescription dose to the tumour are later found using a penalty-based optimization model. We next extend our grassfire and sphere-packing isocentre selection method to treatments with homogenous dose distributions. Dose homogeneity is required in multi-session plans where a larger volume is treated to account for daily setup errors, and thus large overlaps with surrounding healthy tissue may exist. For multi-session plans, we explicitly consider the healthy tissue overlaps in our algorithms and strategically select many isocentres in adjacent volumes to avoid hotspots. There is also interest in treating patients with continuous couch motion to decrease the total treatment session and increase plan quality. We therefore investigate continuous dose delivery treatment plans for PFX. We present various path selection methods along which the dose is delivered using Hamiltonian paths techniques, and develop mixed-integer and linear approximation models to determine the configuration and duration of the radiation time along the paths. We consider several criteria in our optimization models, including machine speed constraints and movement accuracy, preference for single or multiple paths, and smoothness of movement. Our plans in all proposed approaches are tested on seven clinical cases and can meet or exceed clinical guidelines and usually outperform clinical treatments.

En radiothérapie externe, les simulations des dépôts de dose aux patients sont réalisées sur des systèmesde planification de traitement (SPT) dotés d'algorithmes de calcul qui diffèrent dans leur modélisationdes processus physiques d'interaction des électrons et des photons. Or ces SPT, bien que rapides enclinique, montrent parfois des erreurs significatives aux abords des hétérogénéités du corps humain. Montravail de thèse a consisté à valider le modèle aux moments entropique M1 pour des faisceaux d'électronscliniques. Cet algorithme développé au CELIA dans le cadre de la physique des plasmas repose sur larésolution de l'équation cinétique de transport de Boltzmann linéarisée selon une décomposition auxmoments. M1 nécessite une fermeture du système d'équations basée sur le H-Théorème (maximisationde l'entropie). Les cartographies de dose 1D de faisceaux d'électrons de 9 et 20 MeV issues de M1 ontété comparées à celles issues de codes de référence : macro Monte-Carlo clinique (eMC) et full Monte-Carlo (GEANT-MCNPX) ainsi qu'à des données expérimentales. Les cas tests consistent en des fantômesd'abord homogènes puis de complexité croissante avec insertion d'hétérogéenéités mimant les tissus osseuxet pulmonaire. In fine, le modèle aux moments M1 démontre des propriétés de précision meilleures quecertains algorithmes de type Pencil Beam Kernel encore utilisés cliniquement et proches de celles fourniespar des codes full Monte-Carlo académiques ou macro Monte-Carlo cliniques, même dans les cas testscomplexes retenus. Les performances liées aux temps de calcul de M1 ont été évaluées comme étantmeilleures que celles de codes Monte-Carlo. / In radiotherapy field, dose deposition simulations in patients are performed on Treatment Planning Systems (TPS) equipped with specific algorithms that differ in the way they model the physical interaction processes of electrons and photons. Although those clinical TPS are fast, they show significant discrepancies in the neighbooring of inhomogeneous tissues. My work consisted in validating for clinical electron beams an entropic moments based algorithm called M1. Develelopped in CELIA for warm and dense plasma simulations, M1 relies on the the resolution of the linearized Boltzmann kinetic equation for particles transport according to a moments decomposition. M1 equations system requires a closure based on H-Theorem (entropy maximisation). M1 dose deposition maps of 9 and 20 MeV electron beams simulations were compared to those extracted from reference codes simulations : clinical macro Monte-Carlo (eMC) and full Monte-carlo (GEANT4-MCNPX) codes and from experimental data as well. The different test cases consisted in homogeneous et complex inhomogeneous fantoms with bone and lung inserts. We found that M1 model provided a dose deposition accuracy better than some Pencil Beam Kernel algorithm and close of those furnished by clinical macro and academic full Monte-carlo codes, even in the worst inhomogeneous cases. Time calculation performances were also investigated and found better than the Monte-Carlo codes.