Imaging modalities such as CT, MRI, and SPECT have had a tremendous impact on diagnosis and treatment planning. These imaging techniques have given doctors the capability to visualize 3D anatomy structures of human body and soft tissues while being non-invasive. Unfortunately, the 3D images produced by these modalities often have boundaries between the organs and soft tissues that are difficult to delineate due to low signal to noise ratios and other factors. Image segmentation is employed as a method for differentiating Regions of Interest in these images by creating artificial contours or boundaries in the images. There are many different techniques for performing segmentation and automating these methods is an active area of research, but currently there are no generalized methods for automatic segmentation due to the complexity of the problem. Therefore hand-segmentation is still widely used in the medical community and is the €œGold standard€� by which all other segmentation methods are measured. However, existing manual segmentation techniques have several drawbacks such as being time consuming, introduce slice interpolation errors when segmenting slice-by-slice, and are generally not reproducible. In this thesis, we present a novel semi-automated method for 3D hand-segmentation that uses mesh extraction and surface parameterization to project several 3D meshes to 2D plane . We hypothesize that allowing the user to better view the relationships between neighboring voxels will aid in delineating Regions of Interest resulting in reduced segmentation time, alleviating slice interpolation artifacts, and be more reproducible.

La nécessité de protéger les véhicules de rentrée atmosphérique a conduit à la sélection des matériaux carbonés pour les applications les plus extrêmes [1, 2]. Au vu du coût prohibitif des essais en conditions réelles [1], la simulation permet de prédire et comprendre le comportement du matériau lors de la rentrée atmosphérique. Les modèles de l’ablation du carbone n’ont cessé de se complexifier [1, 3, 4, 5, 6]. Cependant, ils ne donnent pas de descriptions fines de la surface du matériau. C’est pour remédier à cela que plusieurs études ont été lancées au Laboratoire des Composites Thermostructuraux (LCTS) [7, 8]. Le Plasmatron du VKI est le moyen à haut flux de chaleur et à haute vitesse utilisé pour cette thèse [9, 10]. Dans cette étude, des échantillons sont testés sous des flux de chaleur allant jusqu’à 2800 kW/m2. Ces essais sur le composite 3D carbone/carbone (Cf/C) ont mené à des résultats très intéressants sur la morphologie du composite lors de l’ablation. Le four à image d’arc est le moyen à très haut flux de chaleur et à convection nulle utilisé au cours de cette thèse. Des flux de chaleurs de 10 MW m-2 ont pu être appliqués à des échantillons de 3D Cf/C. Les observations MEB couplées à la reconstruction numérique du four ont permis de noter les différences entre l’oxydation et la sublimation du composite. Un code de diffusion/ablation est développé depuis plusieurs années au LCTS [7, 11, 12, 13, 14] : AMA. Il permet de simuler l’ablation d’un matériau en carbone. Ce code a été validé par J. Lachaud sans prise en compte de l’écoulement [7]. Un nouveau couplage avec le code Open-Foam [15] a été développé et validé. Le code AMA a été utilisé afin de déterminer les paramètres influençant la rugosité du 3D Cf/C sous un écoulement laminaire. / Carbonaceous materials have been selected to protect atmospheric re-entry vehicles under the most extremes conditions [1, 2]. Because real flight experiments are very expensive [1], simulation is an important means to predict and understand the material behaviour during atmospheric re-entry. The models of carbonaceous materials ablation have become more and more complex [1, 3, 4, 5, 6]. However, they are not accurate to describe finely the material surface. To find a solution to this problem, several studies have been conducted at the Laboratoire des Composites Thermostructuraux (LCTS) [7, 8]. The Plasmatron of the VKI is the experimental means which has been used in this doctoral thesis. It can reach moderately high heat flux under high speed flow [9, 10]. In this study, several samples have been tested under an heat flux up to 2800 kW/m2. These tests on the 3D carbon/carbon (Cf/C) composite have given very interesting results about the material morphology during ablation. An arc image furnace, which can reach very high heat flux but without important flow field, has been used during this doctoral study. 3D Cf/C samples have been ablated under heat flux up to 10 MW m-2. SEM observations with the help of a numerical simulation of the furnace, have put the differences between oxidation and sublimation forward. A diffusion/ablation code has been developed at LCTS for several years [7, 11, 12, 13, 14] : AMA. This piece of software is designed to simulate the ablation of a carbonaceous material. It has been validated without flow field by J. Lachaud [7]. A new coupling method between AMA and Open-Foam [15] has been developed and validated. Finally, AMA has been used to point out the parameters which drive the 3D Cf/C surface recession under laminar flow field.

Cette thèse associe analyse d’image et modélisation physico-chimique afin de caractériser l’infiltrabilité d’un milieu poreux. Infiltrabilité signifie : « propension d’un milieu poreux à se laisser pénétrer par un fluide apportant un dépôt solide ». Une application est la fabrication de composites à matrice céramique par dépôt chimique en phase gazeuse (CVI). Des études ont montré que l’agencement des fibres d’un matériau composite a un impact sur sa densité finale. Nous proposons d’étudier l’évolution du milieu poreux au cours de l’infiltration pour des architectures complexes. La première étape consiste en la segmentation et la caractérisation de composites déjà densifiés obtenus par micro-tomographie. Les objets à segmenter sont des fibres quasi-cylindriques. Deux outils ont été développés : un estimateur optimal de l’orientation vers l’axe de cylindres, et un algorithme de détection et de caractérisation d’objets quasi-cylindriques. Appliquée aux composites fibreux, cette étape fournit un bloc contenant les fibres. Il constitue le milieu poreux complexe dont on cherche à caractériser l’infiltrabilité. La seconde étape est la modélisation à l’échelle des fibres du procédé CVI. Elle utilise des marcheurs aléatoires, avec une gestion de l’interface du solide par « marching cube simplifié». L’algorithme proposé est novateur car il prend en compte simultanément les réactions chimiques, le transport de gaz en régime raréfié ou continu et l’évolution temporelle de la morphologie d’un milieu poreux. Le couplage des deux étapes permet de comparer le dépôt issu de la segmentation à celui résultant de la simulation dans divers régimes physiques. Il est alors possible d’effectuer une analyse inverse des conditions d’élaboration à partir de la morphologie du dépôt. Les outils proposés permettent aussi de comparer l’infiltrabilité de différentes architectures fibreuses. / This thesis connects image processing and physicochemical modeling to characterize the infiltrability of porous media. Infiltrability means “ability of a porous medium to receive a solid deposit brought by penetration of a carrier fluid”. A practical case is the preparation of ceramic-matrix composites by Chemical Vapor Infiltration (CVI). Various studies have proved that the fiber arrangement in preforms of composite materials affects the density of the material at the final stage. In this work, the morphological evolution of complex 3D porous media during the gas-phase infiltration is studied. The first step consists in the segmentation and characterization X-ray Micro Tomography of the infiltrated composite. The objects to be segmented are quasi cylindrical fibers. Two tools have been developed: an optimal estimator of the orientation toward the axis; and an algorithm to detect and characterize quasi cylindrical objects. Applied on images of fiber-reinforced composites, this approach makes it possible to obtain the block containing the fibers. This block is the complex porous medium used for infiltrability characterization. The second step addresses the fiber-scale modeling of CVI. It is based on random walkers and fluid / solid interface management by a simplified marching cube. Our algorithm is innovative since it handles simultaneously chemical reactions, gas transport in rarefied and continuum regimes, and the morphological evolution of porous structure. By combining these two steps, we can compare the deposit obtained by segmentation to simulated deposits obtained in various physicochemical regimes. This allows performing an inverse analysis of the actual deposition conditions from the morphology of the deposit. The provided computational approach also allows the comparison of different porous textures with respect to their infiltrability.

This thesis discusses possibilities of accelerated 3D scene displaying in a Web browser. In more detail, it deals with WebGL standard and its use in real applications. An application for visualization of volumetric medical data based on JavaScript, WebGL and Three.js library was designed and implemented. Image data are loaded from Google Drive cloud storage. An important part of the application is 3D visualization of the volumetric data based on volume rendering technique called Ray-casting.