Doenças vasculares são um problema mundial, que representa 28% das mortes no mundo e 66% do total de doenças que acometem os brasileiros. Dessa forma, há um grande interesse em pesquisar formas de prevenção e tratamento dessas doenças. Algumas medidas são relevantes no auxílio de diagnóstico, tal como: tamanho médio dos ramos, diâmetro médio das seções transversais dos vasos e padrões de divisão de ramos. Calcular essas medidas de forma manual é uma tarefa demorada e trabalhosa. Assim, esta Tese tem como objetivo, propor um método computacional de rastreamento e extração de atributos em redes vasculares a partir de imagens 3D de angiografia por ressonância magnética e por tomografia computadorizada. Trata-se de uma abordagem de rastreamento e identificação de bifurcações que difere das técnicas anteriores, utilizando a Transformada de Hough para identificar o diâmetro do vaso em cortes transversais num dado ponto ao longo de um vaso sanguíneo. Mais detalhadamente, essa abordagem utiliza um campo vetorial advindo do cálculo de uma matriz formada por derivadas parciais de segunda ordem, obtida da intensidade luminosa da imagem, para identificar a direção de um ramo de vaso. Além disso, durante o processo de rastreamento de um ramo de vaso, são calculados vários descritores de forma com o objetivo de classificar regiões como pertencentes a uma bifurcação ou não. Em adição a estes descritores, desenvolvemos uma nova medida chamada de variância do raio que permite distinguir, bifurcações, não-bifurcações e segmentos de vaso com stents (aparelho metálico usado para aumentar o diâmetro dos vasos). Para a classificação de bifurcações, criamos a medida de bifurcação, que trata-se de uma combinação linear de todos os descritores de forma apresentados neste trabalho. Testes foram realizados para atestar a eficácia da abordagem proposta, utilizando tanto imagens sintéticas quantoimagens reais. Os resultados mostraram que o método é capaz de rastrear 91% de uma rede vascular sintética variando o ponto de inicialização e 76% variando o nível de ruído. Também foi observado por meio de testes que o método proposto consegue rastrear vasos e identificar bifurcações em imagens reais sem avaliação numérica. Essa abordagem permite a extração da relação hierárquica entre os ramos em uma rede vascular e a extração do padrão de divisão dos vasos, o que contribui sobremaneira para o estudo do comportamento do fenômeno da angiogênese e no auxílio no diagnóstico de anomalias vasculares. / Vascular diseases are a main health problem, representing 28% of deaths worldwide and 66% of all diseases affecting the Brazilian population. Thus, it is important that researches in prevention and treatment of this type of disease increase. Moreover, there are several demands, such as computational tools capable of analyzing and extracting attributes from non-invasive images. The scope of this work is the analysis and extraction of data from magnetic resonance angiography and computed tomography angiography images by highlighting blood vessels. In this context, this thesis aims the development of a novel computational tracking and feature extraction method for vascular networks from 3D images. Our approach presents the following steps: First, identify the vessel cross-sections along it using the Hough transform. Then, compute a matrix composed of second order partial derivatives of image intensity to identify the direction of the vessel. Perform a feature analysis of the vessel contour to classify the bifurcation point, and finally, identify the direction of the new branch in a bifurcation point. The main contribution of this Thesis is the two new measures developed, called radius ratio and bifurcation measure, the radius ratio is capable to distinguish between a region with bifurcation, stents or without both of them. The bifurcation measure is a linear combination that allows to classify a region as bifurcation or not. Tests were performed in order to verify the proposed approach effectiveness, using both synthetic images and real images. The results showed the method is capable to track 91% of synthetic vascular networks varying the seed point and 76% varying the level of noise. Also, we performed tests in real images and by visual evaluation, we could observed that the proposed method was able to track vessels and identify bifurcations from different parts of the body. This approach allows to calculate, in the future, the density of bifurcations in a vascular network, the distance between them, the stenosis and aneurysms grading and characterize specific vessels. In addition, the vascular networks extraction allows the study of the angiogenesis phenomena and vascular anomalies.

Les images angiographiques sont utilisés pour différentes tâches comme le diagnostique, le suivie de pathologies et la planification d'interventions chirurgicales. Toutefois, en raison du faible ratio signal sur bruit et le contenu complexe des données (informations clairsemées), l'analyse des images angiographiques est une tâche fastidieuse et source d'erreurs. Ces différentes considérations ont motivé le développement de nombreuses techniques d'analyse.Les travaux de cette thèse s'organisent autour de deux axes de recherches : d'une part la segmentation des images angiographiques, et d'autre part la modélisation des réseaux vasculaires cérébraux. En segmentation, l'automatisation induit généralement un coût de calcul élevé, alors que les méthodes interactives sont difficiles à utiliser en raison de la dimension et de la complexité des images. Ces travaux présentent un compromis entre les deux approches, en utilisant le concept de segmentation à base d'exemple. Cette stratégie qui utilise les arbres de coupes de façon non standard,conduit à des résultats satisfaisant, lorsqu'elle est appliqué sur des données d'ARM cérébrales 3D. Les approches existantes, en modélisation, reposent exclusivement sur des informations relatives aux vaisseaux. Ces travaux ont exploré une nouvelle voie, consistant à utiliser à la fois les informations vasculaires et morphologiques ( c-à-d structures cérébrales) pour améliorer la précision et la pertinence des atlas obtenus. Les expériences soulignent des améliorations dans les principales étapes du processus de création de l'atlas impacté par l'utilisation de l'information morphologique. Un exemple d'atlas cérébraux a été réalisé. / Angiographie images are useful data for several tasks, e.g., diagnosis, pathology follow-up or surgery planning. However, due to low SNR (noise,artifacts), and complex semantic content (sparseness), angiographie image analysis is a time consurning and error prone task. These consideration have motivated the development of numerous vesse! filtering, segmentation, or modeling techinques.This thesis is organized around two research areas : the segmentation anù the moùeling. Segmentation of cerebral vascular networks from 3D angiographie data remains a challenge. Automation generally induces a high computational cost and possible errors, white interactive methods are hard to use due to the dimension and the complexity of images. This thesis presents a compromise between both approaches by using the concept of example-based segmentation. This strategy, which uses component-trees in a non-standard fashion, leads to promising results, when applied on cerebral MR angiographie data. The generation of cerebrovascular atlases remains a complex and infrequently considered issue. The existing approaches rely on information exclusively related to the vessels. This thesis investigate a new way, consisting of using both vascular and morphological information (i.e. Cerebral structures) to improve the accuracy and relevance of the obtaines vascular atlases. Experiments emphasize improvments in the main steps of the atlas generation process impacted by the use of the morphological information. An example of cerebrovascular atlas obtained from a dataset of MRAs acquired form different acquisition devices has been provided.

No description available.

Introdução – Pouco se publicou na literatura médica a respeito de projeções angiográficas dedicadas às anatomias especiais. Neste cenário, propomos uma projeção angiográfica para a exposição da bifurcação do tronco da artéria coronária esquerda (TCE) quando o conjunto de projeções rotineiras previamente utilizadas foi ineficaz para exibir com clareza essa região. Pacientes e Métodos – 84 pacientes foram submetidos à projeção Lateral Direita (90-120°) associada a Cranial (30-40°). Estudo de reprodutibilidade, realizado com a participação de dois observadores independentes, julgou a eficácia da projeção proposta. O índice kappa, ajustado para a prevalência e viés (PABAK) com IC de 95%, foi utilizado para demonstrar a intensidade da concordância inter e intraobservador: ≤ 0,20 = ruim, 0,21 – 0,40 = fraca, 0,41 – 0,60 = moderada, 0,61 – 0,80 = substancial e 0,81 – 1,0 = quase perfeita. Resultados – A projeção proposta foi eficaz em 79% dos angiografias com concordância de 0,76 (0,6 – 0,9; P ≤ 0,001). A origem e o segmento proximal da artéria coronária descendente anterior foram expostos em 89% com concordância de 0,86 (0,7 – 1,0; P ≤ 0,001), a origem e o segmento proximal da artéria circunflexa foram expostos em 83% com concordância de 0,72 (0,5 – 1,0; p ≤ 0,001) e a origem e o segmento proximal do ramo intermediário, quando presente, foram expostos em 89% com concordância de 0,79 (0,6 – 1,0; p ≤ 0,001). Conclusão – A projeção proposta é eficaz, segura e reproduzível. Em situações especiais, onde projeções rotineiras falham, essa poderá exibir detalhes importantes da anatomia da bifurcação do TCE à cinecoronariografia ou ser a projeção de trabalho durante a intervenção coronariana percutânea. / Objectives – We propose a 90 – 120° right lateral with 30-40° cranial angiographic view to expose the bifurcation of the left main coronary artery (LMCA) when previously used routine projections were inefficient at clearly showing this region. Background – Little has been published in the medical literature regarding angiographic projections dedicated to special anatomies. Methods – A total of 84 patients were subjected to the proposed projections. A reproducibility study, conducted with the participation of two independent observers, judged the effectiveness of the proposed projection. The Prevalence and Bias Adjusted Kappa (PABAK) index, with a 95% confidence interval (CI), was used to demonstrate the intensity of intra- and inter-observer agreement: ≤ 0.20 = poor, 0.21 – 0.40 = slight, 0.41 – 0.60 = moderate, 0.61 – 0.80 = substantial and 0.81 – 1.0 = almost perfect. Results – The proposed projection was efficient in 79% of the angiograms, with agreement of 0.76 (0.6 – 0.9; P ≤ 0.001). The origin and the proximal segment of the anterior descending coronary artery were exposed in 89% of the angiograms, with agreement of 0.86 (0.7 – 1.0; P ≤ 0.001); the origin and the proximal segment of the circumflex artery were exposed in 83% of the angiograms, with agreement of 0.72 (0.5 – 1.0; P ≤ 0.001); and the origin and the proximal segment of the intermediate branch, when present, were exposed in 89% of the angiograms, with agreement of 0.79 (0.6 – 1.0; P ≤ 0.001). Conclusion – The proposed projection is effective, safe and reproducible. In special situations where routine projections fail, this proposed projection can reveal important details of the anatomy of the bifurcation of the LMCA during conventional / Tese (Doutorado)

La problématique de cette thèse concerne la reconstruction statistique itérative 3D de l'arbre coronaire, à partir d'un nombre très réduit d'angiogrammes coronariens (5 images). Pendant un examen rotationnel d'angiographie RX, seules les projections correspondant à la même phase cardiaque sont sélectionnées afin de vérifier la condition de non variabilité spatio-temporelle de l'objet à reconstruire (reconstruction statique). Le nombre restreint de projections complique cette reconstruction, considérée alors comme un problème inverse mal posé. La résolution d'un tel problème nécessite une procédure de régularisation. Pour ce faire, nous avons opté pour le formalisme bayésien en considérant la reconstruction comme le champ aléatoire maximisant la probabilité a posteriori (MAP), composée d'un terme quadratique de vraisemblance (attache aux données) et un a priori de Gibbs (à priori markovien basé sur une interprétation partielle de l'objet à reconstruire). La maximisation MAP adoptant un algorithme d'optimisation numérique nous a permis d'introduire une contrainte de lissage avec préservation de contours des reconstructions en choisissant adéquatement les fonctions de potentiel associées à l'énergie à priori. Dans ce manuscrit, nous avons discuté en détail des trois principales composantes d'une reconstruction statistique MAP performante, à savoir (1) l'élaboration d'un modèle physique précis du processus d'acquisition, (2) l'adoption d'un modèle à priori approprié et (3) la définition d'un algorithme d'optimisation itératif efficace. Cette discussion nous a conduit à proposer deux algorithmes itératifs MAP, MAP-MNR et MAP-ARTUR-GC, que nous avons testés et évalués sur des données simulées réalistes (données patient issues d'une acquisition CT- 64 multi-barrettes). / The problematic of this thesis concerns the statistical iterative 3D reconstruction of coronary tree from a very few number of coronary angiograms (5 images). During RX rotational angiographic exam, only projections corresponding to the same cardiac phase are selected in order to check the condition of space and time non-variability of the object to reconstruct (static reconstruction). The limited number of projections complicates the reconstruction, considered then as an illness inverse problem. The answer to a similar problem needs a regularization process. To do so, we choose baysian formalism considering the reconstruction as a random field maximizing the posterior probability (MAP), composed by quadratic likelihood terms (attached to data) and Gibbs prior (prior markovian based on a partial interpretation of the object to reconstruct). The MAP maximizing allowed us using a numerical optimization algorithm, to introduce a smoothing constraint and preserve the edges on the reconstruction while choosing wisely the potential functions associated to prior energy. In this paper, we have discussed in details the three components of efficient statistical reconstruction MAP, which are : 1- the construction of precise physical model of acquisition process; 2- the selection of an appropriate prior model; and 3- the definition of an efficient iterative optimization algorithm. This discussion lead us to propose two iterative algorithms MAP, MAP-MNR and MAP-ARTUR-GC, which we have tested and evaluated on realistic simulated data (Patient data from 64-slice CT).