Arnaud Boucher

Dans le monde de la recherche, l’analyse du signal et plus particulièrement d’image, est un domaine très actif, de par la variété des applications existantes, avec des problématiques telles que la compression de données, la vidéo-surveillance ou encore l’analyse d’images médicales pour ne prendre que quelques exemples. Le mémoire s’inscrit dans ce dernier domaine particulièrement actif. Le nombre d’appareils d’acquisition existant ainsi que le nombre de clichés réalisés, entraînent la production d’une masse importante d’informations à traiter par les praticiens. Ces derniers peuvent aujourd’hui être assistés par l’outil informatique. Dans cette thèse, l’objectif est l’élaboration d’un système d’aide au diagnostic, fondé sur l’analyse conjointe, et donc la comparaison d’images médicales. Notre approche permet de détecter des évolutions, ou des tissus aberrants dans un ensemble donné, plutôt que de tenter de caractériser, avec un très fort a priori, le type de tissu cherché.Cette problématique permet d’appréhender un aspect de l’analyse du dossier médical d’un patient effectuée par les experts qui est l’étude d’un dossier à travers le suivi des évolutions. Cette tâche n’est pas aisée à automatiser. L’œil humain effectue quasi-automatiquement des traitements qu’il faut reproduire. Avant de comparer des régions présentes sur deux images, il faut déterminer où se situent ces zones dans les clichés. Toute comparaison automatisée de signaux nécessite une phase de recalage, un alignement des composantes présentes sur les clichés afin qu’elles occupent la même position sur les deux images. Cette opération ne permet pas, dans le cadre d’images médicales, d’obtenir un alignement parfait des tissus en tous points, elle ne peut que minimiser les écarts entre tissus. La projection d’une réalité 3D sur une image 2D entraîne des différences liées à l’orientation de la prise de vue, et ne permet pas d’analyser une paire de clichés par une simple différence entre images. Différentes structurations des clichés ainsi que différents champs de déformation sont ici élaborés afin de recaler les images de manière efficace.Après avoir minimisé les différences entre les positions sur les clichés, l’analyse de l’évolution des tissus n’est pas menée au niveau des pixels, mais à celui des tissus eux-mêmes, comme le ferait un praticien. Afin de traiter les clichés en suivant cette logique, les images numériques sont réinterprétées, non plus en pixels de différentes luminosités, mais en motifs représentatifs de l’ensemble de l’image, permettant une nouvelle décomposition des clichés, une décomposition parcimonieuse. L’atout d’une telle représentation est qu’elle permet de mettre en lumière un autre aspect du signal, et d’analyser sous un angle nouveau, les informations nécessaires à l’aide au diagnostic.Cette thèse a été effectuée au sein du laboratoire LIPADE de l’Université Paris Descartes (équipe SIP, spécialisée en analyse d’images) en collaboration avec la Société Fenics (concepteur de stations d’aide au diagnostic pour l’analyse de mammographies) dans le cadre d’un contrat Cifre.

A thoracic aortic aneurysm is an abnormal dilatation of the aorta that can progress and lead to rupture. The decision to conduct surgery is made by considering the maximum diameter, but it is now well known that this metric alone is not completely reliable. The advent of 4D flow magnetic resonance imaging has allowed for the calculation of new biomarkers for the study of aortic diseases, such as wall shear stress. However, the calculation of these biomarkers requires the precise segmentation of the aorta during all phases of the cardiac cycle. The objective of this work was to compare two different methods for automatically segmenting the thoracic aorta in the systolic phase using 4D flow MRI. The first method is based on a level set framework and uses the velocity field in addition to 3D phase contrast magnetic resonance imaging. The second method is a U-Net-like approach that is only applied to magnitude images from 4D flow MRI. The used dataset was composed of 36 exams from different patients, with ground truth data for the systolic phase of the cardiac cycle. The comparison was performed based on selected metrics, such as the Dice similarity coefficient (DSC) and Hausdorf distance (HD), for the whole aorta and also three aortic regions. Wall shear stress was also assessed and the maximum wall shear stress values were used for comparison. The U-Net-based approach provided statistically better results for the 3D segmentation of the aorta, with a DSC of 0.92 ± 0.02 vs. 0.86 ± 0.5 and an HD of 21.49 ± 24.8 mm vs. 35.79 ± 31.33 mm for the whole aorta. The absolute difference between the wall shear stress and ground truth slightly favored the level set method, but not significantly (0.754 ± 1.07 Pa vs. 0.737 ± 0.79 Pa). The results showed that the deep learning-based method should be considered for the segmentation of all time steps in order to evaluate biomarkers based on 4D flow MRI.

Ascending aortic aneurysm is a pathology that is important to be supervised and treated. During the years the aorta dilates, it becomes stiff, and its elastic properties decrease. In some cases, the aortic wall can rupture leading to aortic dissection with a high mortality rate. The main reference standard to measure when the patient needs to undertake surgery is the aortic diameter. However, the aortic diameter was shown not to be sufficient to predict aortic dissection, implying other characteristics should be considered. Therefore, the main objective of this work is to assess in-vivo the elastic properties of four different quadrants of the ascending aorta and compare the results with equivalent properties obtained ex-vivo. The database consists of 73 cine-MRI sequences of thoracic aorta acquired in axial orientation at the level of the pulmonary trunk. All the patients have dilated aorta and surgery is required. The exams were acquired just prior to surgery, each consisting of 30 slices on average across the cardiac cycle. Multiple deep learning architectures have been explored with different hyperparameters and settings to automatically segment the contour of the aorta on each image and then automatically calculate the aortic compliance. A semantic segmentation U-Net network outperforms the rest explored networks with a Dice score of 98.09% (±0.96%) and a Hausdorff distance of 4.88 mm (±1.70 mm). Local aortic compliance and local aortic wall strain were calculated from the segmented surfaces for each quadrant and then compared with elastic properties obtained ex-vivo. Good agreement was observed between Young’s modulus and in-vivo strain. Our results suggest that the lateral and posterior quadrants are the stiffest. In contrast, the medial and anterior quadrants have the lowest aortic stiffness. The in-vivo stiffness tendency agrees with the values obtained ex-vivo. We can conclude that our automatic segmentation method is robust and compatible with clinical practice (thanks to a graphical user interface), while the in-vivo elastic properties are reliable and compatible with the ex-vivo ones.

Association of quadricuspid aortic valve (QAV) with ascending aortic aneurysms (AsAA) is rare. A 63-year-old female with hypertension was found (on MRI) to have an ascending aortic aneurysm (52 mm in maximum diameter) and dilatation at the level of the sinotubular junction (38 mm in diameter) associated with quadricuspid aortic valve. An ascending aortic wall replacement surgery was performed. In this study, we focus on the behavior of the aorta associated with QAV considering the in vitro biomechanical characteristics and histology. The properties of QAV are closer to bicuspid aortic valve than tricuspid aortic valve, but with higher wall thickness.

Accurate segmentation of the myocardial scar may supply relevant advancements in predicting and controlling deadly ventricular arrhythmias in subjects with cardiovascular disease. In this paper, we propose the architecture of inclusion and classification of prior information U-Net (ICPIU-Net) to efficiently segment the left ventricle (LV) myocardium, myocardial infarction (MI), and microvascular-obstructed (MVO) tissues from late gadolinium enhancement magnetic resonance (LGE-MR) images. Our approach was developed using two subnets cascaded to first segment the LV cavity and myocardium. Then, we used inclusion and classification constraint networks to improve the resulting segmentation of the diseased regions within the pre-segmented LV myocardium. This network incorporates the inclusion and classification information of the LGE-MRI to maintain topological constraints of pathological areas. In the testing stage, the outputs of each segmentation network obtained with specific estimated parameters from training were fused using the majority voting technique for the final label prediction of each voxel in the LGE-MR image. The proposed method was validated by comparing its results to manual drawings by experts from 50 LGE-MR images. Importantly, compared to various deep learning-based methods participating in the EMIDEC challenge, the results of our approach have a more significant agreement with manual contouring in segmenting myocardial diseases.