It is estimated that more than 60% of cancer patients will receive radiotherapy (RT). Medical images acquired from different imaging modalities are used to guide the entire RT process from the initial treatment plan to fractionated radiation delivery. Accurate identification of the gross tumor volume (GTV) on computed tomography (CT), acquired at different time points, is crucial for the success of RT. In addition, complementary information from magnetic resonance imaging (MRI), positron emission tomography (PET), cone-beam computed tomography (CBCT) and electronic portal imaging device (EPID) is often used to obtain better definition of the target, track disease progression and update the radiotherapy plan. However, identifying tumor volumes on medical image data requires significant clinical experience and is extremely time consuming. Computer-based methods have the potential to assist with this task and improve radiotherapy. In this thesis a method was developed for automatically identifying the tumor volume on medical images. The method consists of three main parts: (1) a novel rigid image registration method based on scale invariant feature transform (SIFT) and mutual information (MI); (2) a non-rigid registration (deformable registration) method based on the cubic B-spline and a novel similarity function; (3) a gradient-based level set method that used the registered information as prior knowledge for further segmentation to detect changes in the patient from disease progression or regression and to account for the time difference between image acquisition. Validation was carried out by a clinician and by using objective methods that measure the similarity between the anatomy defined by a clinician and by the method proposed. With this automatic approach it was possible to identify the tumor volume on different images acquired at different time points in the radiotherapy workflow. Specifically, for lung cancer a mean error of 3.9% was found; clinically acceptable results were found for 12 of the 14 prostate cancer cases; and a similarity of 84.44% was achieved for the nasal cancer data. This framework has the potential ability to track the shape variation of tumor volumes over time, and in response to radiotherapy, and could therefore, with more validation, be used for adaptive radiotherapy.

In recent years, the impressive progress performed in imaging, computational and technological fields have made possible the emergence of image-guided radiation therapy (IGRT) and adaptive radiation therapy (ART). The accuracy in radiation dose delivery reached by IMRT offers the possibility to increase locoregional dose-intensity, potentially overcoming the poor tumor control achieved by standard approaches. However, before implementing such a technique in clinical routine, a particular attention has to be paid at the target volumes definition and delineation procedures to avoid inadequate dosage to TVs/OARs. In head and neck squamous cell carcinoma (HNSCC), the GTV is typically defined on CT acquired prior to treatment. However, providing functional information about the tumor, FDG-PET might advantageously complete the classical CT-Scan to better define the TVs. Similarly, re-imaging the tumor with optimal imaging modality might account for the constantly changing anatomy and tumor shape occurring during the course of fractionated radiotherapy. Integrating this information into the treatment planning might ultimately lead to a much tighter dose distribution. From a methodological point of view, the delineation of TVs on anatomical or functional images is not a trivial task. Firstly, the poor soft tissue contrast provided by CT comes out of large interobserver variability in GTV delineation. In this regard, we showed that the use of consistent delineation guidelines significantly improved consistency between observers, either with CT and with MRI. Secondly, the intrinsic characteristics of PET images, including the blur effect and the high level of noise, make the detection of the tumor edges arduous. In this context, we developed specific image restoration tools, i.e. edge-preserving filters for denoising, and deconvolution algorithms for deblurring. This procedure restores the image quality, allowing the use of gradient-based segmentation techniques. This method was validated on phantom and patient images, and proved to be more accurate and reliable than threshold-based methods. Using these segmentation methods, we proved that GTVs significantly shrunk during radiotherapy in patients with HNSCC, whatever the imaging modality used (MRI, CT, FDG-PET). No clinically significant difference was found between CT and MRI, while FDG-PET provided significantly smaller volumes than those based on anatomical imaging. Refining the target volume delineation by means of functional and sequential imaging ultimately led to more optimal dose distribution to TVs with subsequent soft tissue sparing. In conclusion, we demonstrated that a multi-modality-based adaptive planning is feasible in HN tumors and potentially opens new avenues for dose escalation strategies. As a high level of accuracy is required by such approach, the delineation of TVs however requires a special care.

<p>The daily uncertainty concerning tumor localization is one of the major problems during the course of radiation therapy. Image guided-radiation therapy (IGRT) can be used to improve the localization and adjustment of the planning target volume. The aim of this work was to evaluate both the IGRT technique used for prostate cancer patients at the department of the Karolinska University Hospital and an alternative on-line adaptive radiation therapy (ART) method with an On-Board Imager (OBI).</p><p>In the first part of the thesis 2D and 3D image registration with an OBI were compared. Ten prostate cancer patients were involved in the analyses. Two different statistical tests were used to determine significant systematic deviations between the two methods. The second part concerns daily dose verifications and dose plan reoptimization of one intensity modulated radiation therapy (IMRT) prostate cancer patient treated with IGRT. The study was based on cone-beam computed tomography (CBCT) images acquired at 6 different treatment fractions. The risk of developing late rectal and bladder toxicity was quantified using normal tissue complication probability (NTCP) calculations. Additional measurements on an Alderson phantom were performed to verify the accuracy of using the CBCT images for dose calculations.</p><p>A statistically significant difference between the 2D-2D and the 3D-3D match applications could be observed in lateral and longitudinal direction. However, the effect differed among the patients. The phantom measurements showed small dose deviations between the CT and CBCT image, with a mean dose increase to the prostate and seminal vesicles (SV) of 2.5 %. The daily dose to the prostate and SV of the IMRT patient showed to be satisfactory. The daily dose to the rectum did not exceed the prescribed rectal dose except at one treatment fraction and the highest risk of developing late rectal toxicity was about 10.4 %. Large daily bladder dose variations were observed and at two treatment fractions the bladder dose restrictions were exceeded. With a reoptimization process of the dose plan, the dose to the bladder could be reduced while conserving the dose to the target.</p><p>This work shows that for these specific patient cases appropriate doses to the prostate and SV can be delivered with IGRT. However, introducing a suitable ART method could lead to a reduction of inter-fractional rectal and bladder dose variations.</p>

The daily uncertainty concerning tumor localization is one of the major problems during the course of radiation therapy. Image guided-radiation therapy (IGRT) can be used to improve the localization and adjustment of the planning target volume. The aim of this work was to evaluate both the IGRT technique used for prostate cancer patients at the department of the Karolinska University Hospital and an alternative on-line adaptive radiation therapy (ART) method with an On-Board Imager (OBI). In the first part of the thesis 2D and 3D image registration with an OBI were compared. Ten prostate cancer patients were involved in the analyses. Two different statistical tests were used to determine significant systematic deviations between the two methods. The second part concerns daily dose verifications and dose plan reoptimization of one intensity modulated radiation therapy (IMRT) prostate cancer patient treated with IGRT. The study was based on cone-beam computed tomography (CBCT) images acquired at 6 different treatment fractions. The risk of developing late rectal and bladder toxicity was quantified using normal tissue complication probability (NTCP) calculations. Additional measurements on an Alderson phantom were performed to verify the accuracy of using the CBCT images for dose calculations. A statistically significant difference between the 2D-2D and the 3D-3D match applications could be observed in lateral and longitudinal direction. However, the effect differed among the patients. The phantom measurements showed small dose deviations between the CT and CBCT image, with a mean dose increase to the prostate and seminal vesicles (SV) of 2.5 %. The daily dose to the prostate and SV of the IMRT patient showed to be satisfactory. The daily dose to the rectum did not exceed the prescribed rectal dose except at one treatment fraction and the highest risk of developing late rectal toxicity was about 10.4 %. Large daily bladder dose variations were observed and at two treatment fractions the bladder dose restrictions were exceeded. With a reoptimization process of the dose plan, the dose to the bladder could be reduced while conserving the dose to the target. This work shows that for these specific patient cases appropriate doses to the prostate and SV can be delivered with IGRT. However, introducing a suitable ART method could lead to a reduction of inter-fractional rectal and bladder dose variations.

Anatomical changes can have significant clinical impact during head and neck radiotherapy. Adaptive radiotherapy (ART) may be applied to account for such changes. Implementation of ART to alter dose delivery requires deformable image registration (DIR) to assess 3D deformations. This study evaluates the performance and accuracy of a commercial DIR system for clinical applications. The investigations in this project were carried out using images of induced changes in two standard radiotherapy phantoms (RANDO® and CIRS®) and one in-house built phantom. CT image data before and after deformation of the phantoms were processed using Eclipse / SmartAdapt® v.10 system employing a Demons-based algorithm. A DIR protocol was designed, and algorithm performance was assessed quantitatively, using volume analysis and the Dice Similarity Index (DSI), and also evaluated qualitatively. In addition, algorithm performance was assessed for 5 head and neck cancer patients using clinical CT images. Each original planning CT image containing contours of 10 volumes of interest including treatment target volumes and organs at risk was deformed to match a second CT image acquired during the course of the treatment. The original structures were deformed, copied onto the target image and compared to reference contours drawn by 3 radiation oncologists. Phantom investigations gave varied results with average DSI scores ranging from 0.69 to 0.93, with an overall average of 0.86 ± 0.08. These quantitative results were reflected qualitatively, with generally accurate matching between reference and DIR-generated structures. Although air gaps in the phantoms compromised algorithm performance and gave rise to physically aberrant results. Clinical results were generally better with a DSI range of 0.75-0.99 and an overall average of 0.89 ± 0.05, suggesting high DIR accuracy. Qualitatively, some minor contour deformations were noted, as well as artefacts in the axial direction that were due to the CT slice resolution (3 mm) that was used to scan the patients. In addition, contour propagation between images using DIR reduced the time required by physicians to contour the images of head and neck cancer patients by ~47%. This study demonstrated that deformable image registration using a Modified Demons algorithm yields clinically acceptable results and time-saving benefits in contouring that improve clinical workflow. The study also showed that it is feasible to incorporate deformable image registration as part of an adaptive radiotherapy strategy for head and neck cancer, provided further studies are designed to carry out accurate and verifiable dose deformation.

Anatomic changes occur throughout head and neck radiotherapy, and a new treatment plan is often required to mitigate the resulting changes in delivered dose to key structures. This process is known as adaptive radiotherapy (ART), and can be labour-intensive. The aim of this thesis is to optimise ART, addressing some of the technical and clinical challenges facing its routine clinical implementation. Optimising the frequency and timing of adaptive replanning is important, and it has been shown here that intensity modulated radiotherapy (IMRT) and volumetric modulated arc therapy (VMAT) are equally robust to weight loss during head and neck radiotherapy. Plan adaptation strategies that have previously been developed for IMRT are therefore applicable to VMAT.Contour propagation is an important component of ART, and it is essential to ensure that propagated contours are accurate. A method for assessing the suitability of a metric for measuring automatic segmentation accuracy has been developed and applied to the head and neck. For the parotids and larynx, metrics based on surface agreement were better than the commonly used Dice similarity coefficient. By establishing a consensus on which metrics should be used to assess segmentation accuracy, comparison of different algorithms is more objective and should lead to more accurate automatic segmentation. A novel method of assessing contour propagation accuracy on a patient-specific basis has also been developed. This was demonstrated on a cohort of head and neck patients and shows potential as a tool for identifying propagated contours that are subject to a high degree of uncertainty. This is a novel tool that will increase the efficiency of automatic segmentation and, therefore, ART.Optimum ART requires consideration of different radiotherapy-related toxicities, and image-based data mining is a powerful technique for spatially localising dose-response relationships. Correction for multiple comparisons through permutation testing is essential, but has so far only been applied to categorical data. A novel method has been developed for performing permutation testing and image-based data mining with a continuously variable clinical endpoint. Application to trismus for head and neck radiotherapy identified a region with a dose-response relationship in the ipsilateral masseter. Sparing this structure during radiotherapy should reduce the severity of radiation-induced trismus. ART mitigates the dosimetric effects of anatomic changes, and this thesis has addressed technical and clinical challenges that have so far limited its clinical implementation. Detailed knowledge of dose-response relationships will enable selection of patients for ART based on potential clinical benefit, and accurate contour propagation will make ART more efficient, facilitating its routine implementation.

Objectifs : Notre travail avait pour objectifs (i) d’évaluer le bénéfice dosimétrique et de prédire le bénéfice clinique d’une radiothérapie adaptative pour des cancers des voies aéro‐digestives supérieures, à la fois en termes de toxicité et de contrôle local, (ii) d’identifier les patients bons candidats à une stratégie de radiothérapie adaptative, et (iii) d’identifier le meilleur schéma de radiothérapie adaptative pour épargner les parotides. Matériels et méthodes : Le bénéfice dosimétrique a été évalué en utilisant les données de patients inclus dans une étude de phase III évaluant le bénéfice clinique d’une radiothérapie adaptative. La dose cumulée sans et avec radiothérapie adaptative a été estimée par des méthodes de recalage élastique. Une évaluation des différents algorithmes de recalage a été faite à la fois en termes de précision spatiale et d’impact sur la dose estimée. Des modèles de prédiction du risque de surdosage ont été développés en utilisant des modèles linéaires généralisées mixtes et une validation croisée par leave‐one‐out. L’évaluation de différents schémas de radiothérapie adaptative (en termes de fréquence et de nombre) a été réalisée en se basant sur l’épargne des parotides. La valeur prédictive de paramètres quantitatifs issus de la TEP a été évaluée à travers une revue systématique de la littérature. La valeur prédictive de paramètres intensité issue de la TEP a été analysée dans 2 cohortes indépendantes. Résultats : Nos travaux ont confirmé qu’en l’absence de radiothérapie adaptative pour des cancers des VADS, il existe un risque de surdosage des parotides de plus de 2 Gy pour les 2/3 des patients. Il s’y associe un risque de sous dosage de la tumeur de plus de 1 Gy pour 50 % des patients. Une radiothérapie adaptative permet de corriger à la fois le surdosage des parotides (bénéfice clinique estimée de 10 % de diminution du risque de xérostomie) et le sous dosage de la tumeur. Basés sur des paramètres issus de la planification et de la première semaine de traitement, des modèles de prédiction du risque de sur dosage des parotides ou de sous dosage de la tumeur ont été développés. Les paramètres TEP prédictifs du risque de récidive ont été identifiés. Un nomogramme a pu être développé et validé dans une 2nd cohorte de patients. Conclusion : Nos travaux confirment le bénéfice d’une radiothérapie adaptative pour épargner les parotides et maintenir la couverture tumorale. Ce bénéfice dosimétrique devrait permettre une diminution de la toxicité et une amélioration du contrôle local. Des paramètres anatomiques et dosimétriques simples permettent l’identification des patients à risque de surdosage des parotides ou de sous dosage de la tumeur. L’utilisation de la TEP permet d’identifier précocement les patients à haut risque de récidive, candidats potentiels à une intensification thérapeutique. Ces résultats justifient la poursuite des travaux sur une cohorte de patients plus importante, idéalement dans le cadre d’études cliniques de phase III. / Objectifs: The aims of this work were (i) to evaluate the dosimetric benefit and to predict the clinical benefit of adaptive radiotherapy for head and neck cancer, regarding both toxicities and local control, (ii) to identify patients whose good candidate for an adaptive strategy, and (iii) to identify the best adaptive strategy to spare the parotid glands. Materials and methods: The dosimetric benefit was assessed using data from a phase III study evaluating the clinical benefit of an adaptive radiotherapy. Cumulated dose with and without adaptive was estimated using deformable image registration. Different methods of deformable image registration were evaluated regarding both spatial and dose estimation accuracy. Predictive model of the risk of parotid gland overdose was computed using generalized linear mixed model and cross validation by leave‐one‐out. The dosimetric benefit of numerous replanning strategies, defined by various numbers and timing of replanning, with regard to parotid gland sparing, was quantified. We performed a systematic review to evaluate the predictive value of quantitative PET parameters. The predictive value of PET intensity parameters was assessed using two independent cohorts of patients. Résultats: Without adaptive radiotherapy, 65% of the patients had a PG overdose of more than 2 Gy and 50% of the patients had a tumor underdose of more than 1 Gy. Adaptive radiotherapy allows to correct both parotid gland overdose and tumor underdose. Based on parameters calculated at the planning and at the first week of treatment, predictive models of PG overdose and tumor underdose were computed. PET parameters correlated with overall survival were identified. Using two independent cohorts of patients, a nomogram to predict survival was build and externally validated. Conclusion: Our studies showed the benefit of adaptive radiotherapy to spare the parotid glands while increasing tumor coverage. These benefits should allow to decrease the toxicities while increasing local control. Early anatomical and dosimetric parameters allow identifying patients at risk of tumor underdose or parotid gland overdose. PET performed before the treatment allows identifying patients with a high‐risk of locoregional failure and death, potentially candidates for treatment. These results justify further studies on a larger cohort of patients, ideally in phase III clinical trials.

Radiotherapy has become an ever more successful treatment option for cancer.Advances in imaging protocols combined with precise therapy devices suchas linear accelerators contribute towards millimeter precision of treatmentdelivery with far fewer side effects. The ultimate goal of radiotherapy is tomaximize tumor control while minimizing adverse effects to healthy tissues,more importantly organs at risk surrounding the tumor. External beamradiotherapy is currently on the brink of breaking a new frontier: MagneticResonance Imaging (MRI) guided tumor tracking. Here, a combined linearaccelerator and MRI system can be used to treat and follow the tumor duringirradiation, called Real-time Adaptive Radiotherapy (ART). Tailoring of thebeam shape, by means of the Multi-leaf Collimator (MLC) on the fly has thepotential to complete a fully automated radiotherapy process. Recent advances in Reinforcement Learning (RL), a sub field of artificialintelligence has pushed the frontiers further in sequential decision making processesfurther in various fields. In a MLC tracking scenario, we hypothesizethat an RL agent trained on real-time tumor delineations and dose informationcould fulfill a specified dosimetric criteria on the fly over the moving target.To investigate the feasibility of RL for MLC tracking further: we designeda simulator, devised an appropriate RL framework and interfaced them to aDeep Q-Network (DQN) algorithm. Our results demonstrate the feasibility of employing RL for MLC trackingalong with numerous design choices that need to be considered while developingsuch a system. We believe to have taken the first step to bridge MLCtracking and RL by proposing a closed loop solution using dose information.

La radiothérapie moderne combine les techniques complexes et les traitements personnalisés, avec le risque que certaines évolutions et erreurs ayant lieu au cours de traitement passent inaperçues. Ces aléas peuvent entraîner des conséquences graves pour la santé du patient. Dans cette perspective, nous avons étudié le potentiel d'un système de dosimétrie in vivo de transit pour le suivi continu du patient et, par conséquent, la radiothérapie adaptative. L'expérience clinique et des tests de faisabilité ont permis de définir les axes de travail principaux: l'automatisation et la simplification du procédé d'analyse des contrôles. Les développements incluent la création d'une bibliothèque de données standard et une série d'analyses de causes racines, permettant ainsi de renforcer la précision du système, d'améliorer l'automatisation de sa mise en place et d'identifier des pistes pour une analyse efficace des résultats et pour la création d'outils supplémentaires facilitant le suivi et l'adaptation du traitement en routine clinique. / Modern radiation therapy combines complex techniques and personalized treatments, with the risk that certain evolutions and errors occurring during the course of the treatment might go unnoticed. These fluctuations may cause great damage to the health of the patient. In this perspective, we worked on the potential of a transit in vivo dosimetry system for continuous monitoring of the patient and, hereafter, adaptive radiotherapy. Our clinical experience and feasibility testing determined the main lines of work : automatization and simplification of the results analysis method. The developments included the creation of a golden data library and a series of root cause analyzes, allowing us to strengthen the accuracy of the system, to enhance the automatization of the setup and to identify tracks for an efficient analysis of the results and for the creation of additional analytical tools to facilitate the monitoring and adaptation of the treatments in clinical routine.

En radiothérapie externe, le traitement est administré au patient avec une séance chaque jour en utilisant le traitement établi en amont et répété jusqu’à la fin du traitement. La planification du traitement est établie par un système de planification du traitement (TPS) qui utilise des informations basées sur l’anatomie du patient (CT, IRM, TEP) et un calcul de dose. L'intégration aux unités de traitement des dispositifs d´ imagerie embarquée peuvent être utilisées pour vérifier la position du patient et son anatomie (dispositif d’imagerie Portal Électronique - EPID-, tomodensitométrie à faisceau conique - CBCT -). L'objectif principal est de fournir une solution robuste pour la routine afin de contrôler la dose délivrée au patient avec la dose prévue établie à partir du TPS. La première partie de ce travail consiste à intégrer une comparaison d'images EPID pour chaque fraction par rapport à la fraction planifiée (images CT) en acceptant, rejetant ou ignorant la différence de position et de niveau de dose du patient selon une tolérance seuil définie par l’utilisateur. La deuxième partie consiste à utiliser l´ imagerie volumique embarquée pour reconstruire la dose délivrée fraction par fraction en réalisant un calcul de dose. Avec la dose reçue par le patient jour après jour, sur la base de nouveaux indicateurs, il est possible d’établir une traçabilité permettant d’identifier les déviations et d’évaluer la qualité du traitement.Pour conclure, ces nouveaux indicateurs permettraient une double traçabilité : d'une part l'amélioration du calcul de dose puis l'efficacité du traitement ; d´ autre part, la sécurité du traitement et de la technique utilisée. / In external radiation therapy, the treatment is delivered to the patient with a fraction every day using the treatment established before and repeated until the end of the treatment. Treatment planning is establish using anatomical information of the patient and a dose calculation. New technologies already integrated to the treatment units can be used to check the positioning of the patient and anatomical information (Electronic Portal Imaging Device - EPID-, Cone Beam Computed Tomography - CBCT - ).The main goal is to develop a system to control the dose delivered to the patient from the dose establish with the TPS.The first aim of this work is to integrate portal images to compare and assess the position of the patient at each fraction to the planned one (CT images) by accepting, rejecting or ignoring the patient positioning and dose level discrepancy according to a tolerance threshold defined by the user. The second aim is to integrate on boardimaging to reconstruct the delivered dose using a dose calculation. Following the dose received by the patient day after day, based on new indicators, it can be established a traceability to identify deviation and assess treatment quality.To conclude, these new indicators would allow a double traceability : in one hand the improvement of the calculation and then the efficiency of the treatment ; in another hand the safety of the treatment and the technique used.