Analysis of phase-contrast MR images yields cardiac flow information which can be manipulated to produce accurate segmentations of the aorta. New phase contrast segmentation algorithms are proposed that use mean-based calculations and least mean squared curve fitting techniques. A GPU is used to accelerate these algorithms and it is shown that it is possible to achieve up to a 2760x speedup relative to the CPU computation times. Level sets are applied to a magnitude image, where initial conditions are given by the previous segmentation algorithms. A qualitative comparison of results shows that the algorithm parallelized on the GPU appears to produce the most accurate segmentation. After segmentation, particle trace simulations are run to visualize flow patterns in the aorta. A procedure for the definition of analysis planes is proposed from which virtual particles can be emitted/collected within the vessel, which is useful for future quantification of various flow parameters. / October 2014

La segmentation d'image avec a priori de forme a fait l'objet d'une attention particulière ces dernières années. La plupart des travaux existants reposent sur des espaces de formes linéarisés avec de petits modes de déformations autour d'une forme moyenne. Cette approche n'est pertinente que lorsque les formes sont relativement similaires. Dans cette thèse, nous introduisons un nouveau cadre dans lequel il est possible de manipuler des a priori de formes plus généraux. Nous modélisons une catégorie de formes comme une variété de dimension finie, la variété des formes a priori, que nous analysons à l'aide d'échantillons de formes en utilisant des techniques de réduction de dimension telles que les diffusion maps. Un plongement dans un espace réduit est alors appris à partir des échantillons. Cependant, ce modèle ne fournit pas d'opérateur de projection explicite sur la variété sous-jacente et nous nous attaquons à ce problème. Les contributions de ce travail se divisent en trois parties. Tout d'abord, nous proposons différentes solutions au problème des "out-of-sample" et nous définissons trois forces attirantes dirigées vers la variété. 1. Projection vers le point le plus proche; 2. Projection ayant la même valeur de plongement; 3. Projection à valeur de plongement constant. Ensuite, nous introduisons un terme d'a-priori de formes pour les coutours/régions actifs/ves. Un terme d'énergie non-linéaire est alors construit pour attirer les formes vers la variété. Enfin, nous décrivons un cadre variationnel pour le debruitage de variété. Des résultats sur des objets réels tels que des silhouettes de voitures ou des structures anatomiques montrent les possibilités de notre méthode.

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.

Nous présentons dans cette thèse un nouveau modèle statistique de forme et l'utilisons pour la segmentation d'images avec a priori. Ce modèle est représenté par un champ de Markov. Les noeuds du graphe correspondent aux points de contrôle situés sur le contour de la forme géométrique, et les arêtes du graphe représentent les dépendances entre les points de contrôle. La structure du champ de Markov est déterminée à partir d'un ensemble de formes, en utilisant des techniques d'apprentissage de variétés et de groupement non-supervisé. Les contraintes entre les points sont assurées par l'estimation des fonctions de densité de probabilité des longueurs de cordes normalisées. Dans une deuxième étape, nous construisons un algorithme de segmentation qui intègre le modèle statistique de forme, et qui le relie à l'image grâce à un terme région, à travers l'utilisation de diagrammes de Voronoi. Dans cette approche, un contour de forme déformable évolue vers l'objet à segmenter. Nous formulons aussi un algorithme de segmentation basé sur des détecteurs de points d'intérêt, où le terme de régularisation est lié à l'apriori de forme. Dans ce cas, on cherche à faire correspondre le modèle aux meilleurs points candidats extraits de l'image par le détecteur. L'optimisation pour les deux algorithmes est faite en utilisant des méthodes récentes et efficaces. Nous validons notre approche à travers plusieurs jeux de données en 2D et en 3D, pour des applications de vision par ordinateur ainsi que l'analyse d'images médicales.

This thesis describes a probabilistic approach to imagesegmentation and interpretation. The focus of the investigation is the development of a systematic way of combining color, brightness, texture and geometric features extracted from an image to arrive at a consistent interpretation for each pixel in the image. The contribution of this thesis is thus the presentation of a novel framework for the fusion of extracted image features producing a segmentation of an image into relevant regions. Further, a solution to the sub-pixel mixing problem is presented based on solving a probabilistic linear program. This work is specifically aimed at interpreting and digitizing multi-spectral aerial imagery of the Earth's surface. The features of interest for extraction are those of relevance to environmental management, monitoring and protection. The presented algorithms are suitable for use within a larger interpretive system. Some results are presented and contrasted with other techniques. The integration of these algorithms into a larger system is based firmly on a probabilistic methodology and the use of statistical decision theory to accomplish uncertain inference within the visual formalism of a graphical probability model.

The rapid proliferation of 3D volume data, including MRI and CT scans, is prompting the search within computer graphics for more effective volume visualisation techniques. Partially because of the traditional association with medical subjects, concepts borrowed from the domain of scientific illustration show great promise for enriching volume visualisation. This thesis describes the first general system dedicated to creating user-directed, variable-detail, scientific illustrations directly from volume data. In particular, using volume segmentation for explicit abstraction in non-photorealistic volume renderings is a new concept. The unique challenges and opportunities of volume data require rethinking many non-photorealistic algorithms that traditionally operate on polygonal meshes. The resulting 2D images are qualitatively different from but complementary to those normally seen in computer graphics, and inspire an analysis of the various artistic implications of volume models for scientific illustration.

Widespread droughts can have a significant impact on western United States streamflow, but the causes of these events are not fully understood. This dissertation examines streamflow from multiple western US basins and establishes the robust, leading modes of variability in interannual streamflow throughout the past century. I show that approximately 50% of this variability is associated with spatially widespread streamflow anomalies that are statistically independent from streamflow's response to the El Niño-Southern Oscillation (ENSO). The ENSO-teleconnection accounts for approximately 25% of the interannual variability in streamflow, across this network. These atmospheric circulation anomalies associated with the most spatially widespread variability are associated with the Aleutian low and the persistent coastal atmospheric ridge in the Pacific Northwest. I use a watershed segmentation algorithm to explicitly track the position and intensity of these features and compare their variability to the multi-basin streamflow variability. Results show that latitudinal shifts in the coastal atmospheric ridge are more strongly associated with streamflow's north-south dipole response to ENSO variability while more spatially widespread anomalies in streamflow most strongly relate to seasonal changes in the coastal ridge intensity. This likely reflects persistent coastal ridge blocking of cool-season precipitation into western US river basins. I utilize the 35 model runs of the Community Earth System Model Large Ensemble (CESMLE) to determine whether the model ensemble simulates the anomalously strong coastal ridges and extreme widespread wintertime precipitation anomalies found in the observation record. Though there is considerable bias in the CESMLE, the CESMLE runs simulate extremely widespread dry precipitation anomalies with a frequency of approximately one extreme event per century during the historical simulations (1920 - 2005). These extremely widespread dry events correspond significantly with anomalously intense coastal atmospheric ridges. The results from these three papers connect widespread interannual streamflow anomalies in the western US - and especially extremely widespread streamflow droughts - with semi-permanent atmospheric ridge anomalies near the coastal Pacific Northwest. This is important to western US water managers because these widespread events appear to have been a robust feature of the past century. The semi-permanent atmospheric features associated with these widespread dry streamflow anomalies are projected to change position significantly in the next century as a response to global climate change. This may change widespread streamflow anomaly characteristic in the western US, though my results do not show evidence of these changes within the instrument record of last century.

Modern medical imaging devices are able to generate highly detailed three-dimensional (3D) images of the skeleton. Computerized image processing and analysis methods, combined with real-time volume visualization techniques, can greatly facilitate the interpretation of such images and are increasingly used in surgical planning to aid reconstruction of the skeleton after trauma or disease. Two key challenges are to accurately separate (segment) bone structures or cavities of interest from the rest of the image and to interact with the 3D data in an efficient way. This thesis presents efficient and precise interactive methods for segmenting, visualizing, and analysing 3D computed tomography (CT) images of the skeleton. The methods are validated on real CT datasets and are primarily intended to support planning and evaluation of cranio-maxillofacial (CMF) and orthopedic surgery. Two interactive methods for segmenting the orbit (eye-socket) are introduced. The first method implements a deformable model that is guided and fitted to the orbit via haptic 3D interaction, whereas the second method implements a user-steered volumetric brush that uses distance and gradient information to find exact object boundaries. The thesis also presents a semi-automatic method for measuring 3D angulation changes in wrist fractures. The fractured bone is extracted with interactive mesh segmentation, and the angulation is determined with a technique based on surface registration and RANSAC. Lastly, the thesis presents an interactive and intuitive tool for segmenting individual bones and bone fragments. This type of segmentation is essential for virtual surgery planning, but takes several hours to perform with conventional manual methods. The presented tool combines GPU-accelerated random walks segmentation with direct volume rendering and interactive 3D texture painting to enable quick marking and separation of bone structures. It enables the user to produce an accurate segmentation within a few minutes, thereby removing a major bottleneck in the planning procedure.

En traitement du signal audio, les plosives sont des sons de parole très importants au regard de l’intelligibilité et de la qualité. Les plosives sont cependant difficiles à modéliser à l’aide des techniques usuelles (prédiction linéaire et codage par transformée), à cause de leur dynamique propre importante et à cause de leur nature non prédictible. Cette étude présente un exemple de système complet capable de détecter, segmenter, et altérer les plosives dans un flux de parole. Ce système est utilisé afin de vérifier la validité de l’hypothèse suivante : La phase d’éclatement (de burst) des plosives peut être mise à zéro, de façon perceptuellement équivalente. L’impact sur la qualité subjective de cette transformation est évalué sur une banque de phrases enregistrées. Les résultats de cette altération hautement destructive des signaux tendent à montrer que l’impact perceptuel est mineur. Les implications de ces résultats pour le codage de la parole sont abordées.

Medical volume segmentation got the attraction of many researchers; therefore, many techniques have been implemented in terms of medical imaging including segmentations and other imaging processes. This research focuses on an implementation of segmentation system which uses several techniques together or on their own to segment medical volumes, the system takes a stack of 2D slices or a full 3D volumes acquired from medical scanners as a data input. Two main approaches have been implemented in this research for segmenting medical volume which are multi-resolution analysis and statistical modeling. Multi-resolution analysis has been mainly employed in this research for extracting the features. Higher dimensions of discontinuity (line or curve singularity) have been extracted in medical images using a modified multi-resolution analysis transforms such as ridgelet and curvelet transforms. The second implemented approach in this thesis is the use of statistical modeling in medical image segmentation; Hidden Markov models have been enhanced here to segment medical slices automatically, accurately, reliably and with lossless results. But the problem with using Markov models here is the computational time which is too long. This has been addressed by using feature reduction techniques which has also been implemented in this thesis. Some feature reduction and dimensionality reduction techniques have been used to accelerate the slowest block in the proposed system. This includes Principle Components Analysis, Gaussian Pyramids and other methods. The feature reduction techniques have been employed efficiently with the 3D volume segmentation techniques such as 3D wavelet and 3D Hidden Markov models. The system has been tested and validated using several procedures starting at a comparison with the predefined results, crossing the specialists’ validations, and ending by validating the system using a survey filled by the end users explaining the techniques and the results. This concludes that Markovian models segmentation results has overcome all other techniques in most patients’ cases. Curvelet transform has been also proved promising segmentation results; the end users rate it better than Markovian models due to the long time required with Hidden Markov models.