Дисертації з теми "Echocardiography segmentation"

Thesis (Ph. D.)--Ohio State University, 2004.
Title from first page of PDF file. Document formatted into pages; contains xvii, 151 p.; also includes graphics (some col.). Includes bibliographical references (p. 145-151). Available online via OhioLINK's ETD Center

Thèse doctorat : Images et Systèmes : Villeurbanne, INSA : 2003.
Thèse rédigée en anglais. Résumé en français en début de chaque chapitre. Titre provenant de l'écran-titre. Bibliogr. p. 216-232. Publications de l'auteur p. 214-215.

Heart disease is a major cause of death in western countries. In order to diagnose and monitor heart disease, 3D echocardiography is an important tool, as it provides a fast, relatively low-cost, portable and harmless way of imaging the moving heart. Segmentation of cardiac walls is an indispensable method of obtaining quantitative measures of heart function. However segmentation of ultrasound images has its challenges: image quality is often relatively low and current segmentation methods are often not fast. It is desirable to make the segmentation technique as fast as possible, making quantitative heart function measures available at the time of recording. In this thesis, we test two state-of-the-art fast segmentation techniques to address this issue; furthermore, we develop a novel technique for finding the best segmentation propagation strategy between points of time in a cardiac image sequence. The first fast method is Graph Cuts (GC), an energy minimisation technique that represents the image as a graph. We test this method on static 3D echocardiography to segment the myocardium, varying the importance of the regulariser function. We look at edge measures, position constraints and tissue characterisation and find that GC is relatively fast and accurate. The second fast method is Random Forests (RFos), a discriminative classifier using binary decision trees, used in machine learning. To our knowledge, we are the first to test this method for myocardial segmentation on 2D and 3D static echocardiography. We investigate the number of trees, image features used, some internal parameters, and compare with intensity thresholding. We conclude that RFos are very fast and more accurate than GC segmentation. The static RFo method is subsequently applied to all time frames. We describe a novel optical flow based propagation technique that improves the static results by propagating the results from well-performing time frames to less-performing frames. We describe a learning algorithm that learns for each frame which propagation strategy is best. Furthermore, we look at the influence of the number of images and of the training set available per tree, and we compare against other methods that use motion information. Finally, we perform the same propagation learning method on the static GC results, concluding that the propagation method improves the static results in this case as well. We compare the dynamic GC results with the dynamic RFo results and find that RFos are more accurate and faster than GC.

Part 1: The bandwidth of the ultrasound Doppler signal is proposed as a classification function of blood and tissue signal in transthoracial echocardiography of the left ventricle. The new echocardiographic mode, Bandwidth Imaging, utilizes the difference in motion between tissue and blood. Specifically, Bandwidth Imaging is the absolute value of the normalized autocorrelation function with lag one. Bandwidth Imaging is therefore linearly dependent on the the square of the bandwidth estimated from the Doppler spectrum. A 2-tap Finite Impulse Response high-pass filter is used prior to autocorrelation calculation to account for the high level of DC clutter noise in the apical regions. Reasonable pulse strategies are discussed and several images of Bandwidth Imaging are included. An in vivo experiment is presented, where the apparent error rate of Bandwidth Imaging is compared with apparent error rate of Second-Harmonic Imaging on 15 healthy men. The apparent error rate is calculated from signal from all myocardial wall segments defined in \cite{Cer02}. The ground truth of the position of the myocardial wall segments is determined by manual tracing of endocardium in Second-Harmonic Imaging. A hypotheses test of Bandwidth Imaging having lower apparent error rate than

Second-Harmonic Imaging is proved for a p-value of 0.94 in 3 segments of end diastole and 1 segment in end systole on non averaged data. When data is averaged by a structural element of 5 radial, 3 lateral and 4 temporal samples, the numbers of segments are increased to 9 in end diastole and to 6 in end systole. These segments are mostly located in apical and anterior wall regions. Further, a global measure GM is defined as the proportion of misclassified area in the regions close to endocardium in an image. The hypothesis test of Second-Harmonic Imaging having lower GM than Bandwidth Imaging is proved for a p-value of 0.94 in the four-chamber view in end systole in any type of averaging. On the other side, the hypothesis test of Bandwidth Imaging having lower GM than Second-Harmonic Imaging is proved for a p-value of 0.94 in long-axis view in end diastole in any type of averaging. Moreover, if images are averaged by the above structural element the test indicates that Bandwidth Imaging has a lower apparent error rate than Second-Harmonic Imaging in all views and times (end diastole or end systole), except in four-chamber view in end systole. This experiment indicates that Bandwidth Imaging can supply additional information for automatic border detection routines on endocardium.

Part 2: Knowledge Based Imaging is suggested as a method to distinguish blood from tissue signal in transthoracial echocardiography. This method utilizes the maximum likelihood function to classify blood and tissue signal. Knowledge Based Imaging uses the same pulse strategy as Bandwidth Imaging, but is significantly more difficult to implement. Therefore, Knowledge Based Imaging and Bandwidth Imaging are compared with Fundamental Imaging by a computer simulation based on a parametric model of the signal. The rate apparent error rate is calculated in any reasonable tissue to blood signal ratio, tissue to white noise ratio and clutter to white noise ratio. Fundamental Imaging classifies well when tissue to blood signal ratio is high and tissue to white noise ratio is higher than clutter to white noise ratio. Knowledge Based Imaging classifies also well in this environment. In addition, Knowledge Based Imaging classifies well whenever blood to white noise ratio is above 30 dB. This is the case, even when clutter to white noise ratio is higher than tissue to white noise ratio and tissue to blood signal ratio is zero. Bandwidth Imaging performs similar to Knowledge Based Imaging, but blood to white noise ratio has to be 20 dB higher for a reasonable classification. Also the highpass filter coefficient prior to Bandwidth Imaging calculation is discussed by the simulations. Some images of different parameter settings of Knowledge Based Imaging are visually compared with Second-Harmonic Imaging, Fundamental Imaging and Bandwidth Imaging. Changing parameters of Knowledge Based Imaging can make the image look similar to both Bandwidth Imaging and Fundamental Imaging.

Mitral valve (MV) disease is responsible for approximately 2,581 deaths and 41,000 hospital discharges each year in the US. Mitral regurgitation (MR), retrograde blood from through the MV, is often an indicator of MV disease. Surgical repair of MVs is preferred over replacement, as it is correlated with better patient quality of life. However, replacement rates are still near 40% because MV surgical repair expertise is not spread across all hospitals. In addition, 15-80% of surgical repair patients have recurrent MR within 10 years. Quantitative patient-specific models could aid these issues by providing less experienced surgeons with additional information before surgery and a quantitative map of patient valve changes after surgery. Real-time 3D echocardiography (RT3DE) can provide high quality 3D images of MVs and has been used to generate quantitative models previously. However, there is not currently an efficient, dynamic, and validated method that is fast enough to use in common practice. To fill this need, a tool to generate quantitative 3D models of mitral valve leaflets from RT3DE in an efficient manner was created. Then an in vitro echocardiography correction scheme was devised and a dynamic, in vitro validation of the tool was performed. The tool demonstrated that it could generate dynamic, complex MV geometry accurately and more efficiently than current methods available. In addition, the ability for mesh interpolation techniques to reduce segmentation time was demonstrated. The tool generated by this study provides a method to quickly and accurately generate MV geometry that could be applied to dynamic patient specific geometry to aid surgical decisions and track patient geometry changes after surgery.

Congenital heart disease affects just under one percentage of all live births [1]. Those defects that manifest themselves as changes to the cardiac chamber volumes are the motivation for the research presented in this thesis. Blood volume measurements in vivo require delineation of the cardiac chambers and manual tracing of foetal cardiac chambers is very time consuming and operator dependent. This thesis presents a multi region based level set snake deformable model applied in both 2D and 3D which can automatically adapt to some extent towards ultrasound noise such as attenuation, speckle and partial occlusion artefacts. The algorithm presented is named Mumford Shah Sarti Collision Detection (MSSCD). The level set methods presented in this thesis have an optional shape prior term for constraining the segmentation by a template registered to the image in the presence of shadowing and heavy noise. When applied to real data in the absence of the template the MSSCD algorithm is initialised from seed primitives placed at the centre of each cardiac chamber. The voxel statistics inside the chamber is determined before evolution. The MSSCD stops at open boundaries between two chambers as the two approaching level set fronts meet. This has significance when determining volumes for all cardiac compartments since cardiac indices assume that each chamber is treated in isolation. Comparison of the segmentation results from the implemented snakes including a previous level set method in the foetal cardiac literature show that in both 2D and 3D on both real and synthetic data, the MSSCD formulation is better suited to these types of data. All the algorithms tested in this thesis are within 2mm error to manually traced segmentation of the foetal cardiac datasets. This corresponds to less than 10% of the length of a foetal heart. In addition to comparison with manual tracings all the amorphous deformable model segmentations in this thesis are validated using a physical phantom. The volume estimation of the phantom by the MSSCD segmentation is to within 13% of the physically determined volume.

L’objectif principal de cette thèse est le développement de techniques de segmentation et de suivi totalement automatisées du ventricule gauche (VG) en RT3DE. Du fait de la nature difficile et complexe des données RT3DE, l’application directe des algorithmes classiques de vision par ordinateur est le plus souvent impossible. Les solutions proposées ont donc été formalisées et implémentées de sorte à satisfaire les contraintes suivantes : elles doivent permettre une analyse complètement automatique (ou presque) et le temps de calcul nécessaire doit être faible afin de pouvoir fonctionner en temps réel pour une utilisation clinique optimale. Dans ce contexte, nous avons donc proposé un nouveau cadre ou les derniers développements en segmentation d’images par ensembles de niveaux peuvent être aisément intégrés, tout en évitant les temps de calcul importants associés à ce type d’algorithmes. La validation clinique de cette approche a été effectuée en deux temps. Tout d’abord, les performances des outils développés ont été évaluées dans un contexte global se focalisant sur l’utilisation en routine clinique. Dans un second temps, la précision de la position estimée du contour du ventricule gauche a été mesurée. Enfin, les méthodes proposées ont été intégrées dans une suite logicielle utilisée à des fins de recherche. Afin de permettre une utilisation quotidienne efficace, des solutions conviviales ont été proposées incluant notamment un outil interactif pour corriger la segmentation du VG
The fundamental goal of the present thesis was the development of automatic strategies for left ventricular (LV) segmentation and tracking in RT3DE data. Given the challenging nature of RT3DE data, classical computer vision algorithms often face complications when applied to ultrasound. Furthermore, the proposed solutions were formalized and built to respect the following requirements: they should allow (nearly) fully automatic analysis and their computational burden should be low, thus enabling real-time processing for optimal online clinical use. With this in mind, we have proposed a novel segmentation framework where the latest developments in level-set-based image segmentation algorithms could be straightforwardly integrated, while avoiding the heavy computational burden often associated with level-set algorithms. Furthermore, a strong validation component was included in order to assess the performance of the proposed algorithms in realistic scenarios comprising clinical data. First, the performance of the developed tools was evaluated from a global perspective, focusing on its use in clinical daily practice. Secondly, also the spatial accuracy of the estimated left ventricular boundaries was assessed. As a final step, we aimed at the integration of the developed methods in an in-house developed software suite used for research purposes. This included user-friendly solutions for efficient daily use, namely user interactive tools to adjust the segmented left ventricular boundaries

A Ecocardiografia continua sendo a técnica de captura de imagens mais promissora, não-invasiva, sem radiação ionizante e de baixo custo para avaliação de condições cardíacas. Porém, é afetada consideravelmente por ruídos do tipo speckle, que são difíceis de serem filtrados. Por isso fez-se necessário fazer a escolha certa entre filtragem e segmentador para a obtenção de resultados melhores na segmentação de estruturas. O objetivo dessa pesquisa foi estudar essa combinação entre filtro e segmentador. Para isso, foi desenvolvido um sistema segmentador, a fim de sistematizar essa avaliação. Foram implementados dois filtros para atenuar o efeito do ruído speckle - Linear Scaling Mean Variance (LSMV) e o filtro de Chitwong - testados em imagens simuladas. Foram simuladas 60 imagens com 300 por 300 pixels, 3 modelos, 4 espessuras e 5 níveis de contrastes diferentes, todas com ruído speckle. Além disso, foram feitos testes com a combinação de filtros. Logo após, foi implementado um algoritmo de conectividade Fuzzy para fazer a segmentação e um sistema avaliador, seguindo os critérios descritos por Loizou, que faz a contagem de verdadeiro-positivos (VP) e falso-positivos (FP). Foi verificado que o filtro LSMV é a melhor opção para segmentação por conectividade Fuzzy. Foram obtidas taxas de VP e FP na ordem de 95% e 5%, respectivamente, e acurácia em torno de 95%. Para imagens ruidosas com alto contraste, aplicando a segmentação sem filtragem, a acurácia obtida foi na ordem de 60%.
The echocardiography is the imaging technique that remains most promising, noninvasive, no ionizing radiation and inexpensive to assess heart conditions. On the other hand, is considerably affected by noises, such as speckle, that are very difficult to be filtered. That is why it is necessary to make the right choice of filter and segmentation method to obtain the best results on image segmentation. The goal was evaluate this filter and segmentation method combination. For that, it was developed a segmentation system, to help the assessment. Two filters were implemented to mitigate the effect of speckle noise Linear Scaling Mean Variance (LSMV) and the filter presented by Chitwong - to be tested in simulated images. We simulated 60 images, with size 300 by 300 pixels, 3 models, 4 thicknesses and 5 different levels of contrast, all with speckle noise. In addition, tests were made with a combination of filters. Furthermore, it was implemented a Fuzzy Connectedness algorithm and an evaluation system, following the criteria described by Loizou, which makes the true positives (TP) and false positives (FP) counting. It was found that the LSMV filter is the best option for Fuzzy Connectedness. We obtained rates of TP and FP of 95% and 5% using LSMV, and accuracy of 95%. Using high contrast noisy images, without filtering, we obtained the accuracy in order of 60%.

This biomedical engineering thesis explores the opportunities and challenges of 2D+t contrast echocardiography for left ventricle functional analysis, both clinically and within a computer vision atlas-based deformable template model framework. A database was created for the experiments in this thesis, with 21 studies of contrast Dobutamine Stress Echo, in all 4 principal planes. The database includes clinical variables, human expert hand-traced myocardial contours and visual scoring. First the problem is studied from a clinical perspective. Quantification of endocardial global and local function using standard measures shows expected values and agreement with human expert visual scoring, but the results are less reliable for myocardial thickening. Next, the problem of segmenting the endocardium with a computer is posed in a standard landmark and atlas-based deformable template model framework. The underlying assumption is that these models can emulate human experts in terms of integrating previous knowledge about the anatomy and physiology with three sources of information from the image: texture, geometry and kinetics. Probabilistic atlases of contrast echocardiography are computed, while noting from histograms at selected anatomical locations that modelling texture with just mean intensity values may be too naive. Intensity analysis together with the clinical results above suggest that lack of external boundary definition may preclude this imaging technique for appropriate measuring of myocardial thickening, while endocardial boundary definition is appropriate for evaluation of wall motion. Geometry is presented in a Principal Component Analysis (PCA) context, highlighting issues about Gaussianity, the correlation and covariance matrices with respect to physiology, and analysing different measures of dimensionality. A popular extension of deformable models ---Active Appearance Models (AAMs)--- is then studied in depth. Contrary to common wisdom, it is contended that using a PCA texture space instead of a fixed atlas is detrimental to segmentation, and that PCA models are not convenient for texture modelling. To integrate kinetics, a novel spatio-temporal model of cardiac contours is proposed. The new explicit model does not require frame interpolation, and it is compared to previous implicit models in terms of approximation error when the shape vector changes from frame to frame or remains constant throughout the cardiac cycle. Finally, the 2D+t atlas-based deformable model segmentation problem is formulated and solved with a gradient descent approach. Experiments using the similarity transformation suggest that segmentation of the whole cardiac volume outperforms segmentation of individual frames. A relatively new approach ---the inverse compositional algorithm--- is shown to decrease running times of the classic Lucas-Kanade algorithm by a factor of 20 to 25, to values that are within real-time processing reach.

Com os avanços tecnológicos das últimas décadas, a ecocardiografia surgiu como uma alternativa de diagnóstico por imagem de relativo baixo custo, que não faz uso de energia ionizante ou radioativa. Recentemente, o advento dos agentes de contraste por microbolhas e dos transdutores matriciais tornou possível a visualização tridimensional da anatomia das artérias coronárias. Neste projeto, é proposta a avaliação de métodos de segmentação capazes de visibilizar as artérias coronárias epicárdicas em Imagens de ecocardiografias tridimensionais com contraste de microbolhas. Esse é o primeiro passo para o desenvolvimento de ferramentas computacionais eficazes e eficientes na assistência não invasiva ao acompanhamento do quadro clínico de pacientes, do diagnóstico ao pós-operatório. Propõe-se, uma metodologia que facilite o acesso às coronárias a partir de imagens de ecocardiografia tridimensionais com aplicação de contraste por microbolhas. Dentre as metodologias estudadas, as técnicas baseadas na teoria Fuzzy Connectedness (FC) foram identificadas como as mais promissoras. Estudou-se, portanto, seis abordagens baseadas nessa teoria, três delas são descritas na literatura (Generalized FC GFC; Relative FC RFC; Dynamic Weighted FC DyWFC) e três proposições originais (Area of Search FC ASFC; Ultrasound-k FC USFC; Guided FC GuFC). Para avaliar a acurácia desses algoritmos, confeccionou-se um conjunto de imagens simuladas, composto por 360 imagens, e selecionou-se um conjunto de imagens de exames reais, composto de 10 imagens reais de pacientes com quadro de Cardiomiopatia Hipertrópica. Para as imagens simuladas, os métodos da literatura alcançaram acurácia de 85,5% para GFC, 89,5% para RFC e 92,0% para DyWFC. Enquanto isso, os métodos propostos alcançaram acurácia de 88,9% para ASFC, 91,7 % para USkFC e 95,2% para GuFC. Para as imagens reais, os métodos convergiram para uma segmentação satisfatória quanto à usabilidade na clínica médica. Esses resultados demonstraram, ainda, o melhor desempenho do método proposto GuFC ante os demais. Dessa forma, ele se torna um candidato para ingressar na etapa de segmentação de uma ferramenta computacional para visibilização das coronárias epicárdicas no futuro
With the technological advances of recent decades, echocardiography has emerged as a relatively low cost imaging diagnostic alternative, that does not use ionizing or radioactive energy. Lately, the advent of microbubble-based contrast agents and array transducers turned possible the visualization of three-dimensional coronary arteries anatomy. The present project proposes to evaluate segmentation methods able to deal with the visualization of the epicardial coronary arteries in microbubble-based three-dimensional echocardiography images. This is the first step towards the development of effective and efficient computational tools for diagnosis and prognosis assistance of cardiac pacient. We propose a methodology to facilitate the access to epicardial coronary arteries in tridimensional echocardiographic images. Among the studied approaches, Fuzzy Connectednessbased segmentation methods were identified as being the most promising. We studied six approaches based on this theory, three of them are described in the literature (Generalized FC GFC; Relative FC RFC; Dynamic Weighted FC DyWFC) and three original contributions (Area of Search FC ASFC; Ultrasound-k FC USFC; Guided FC GuFC). To evaluate the accuracy of these algorithms, a set composed of 360 simulated images were created. We also selected a set of 10 real images, composed of hypertrophic cardiomyopathy patients. For simulated images set, the methods of literature achieved accuracy of 85.5% for GFC, 89,5% for RFC and 92,0% for DyWFC, meanwhile, the proposed method achieved accuracy of 88.9% for ASFC, 91,7 % for USkFC and 95,2% for GuFC. Using the real images set, the methods converged to good results for clinical purposes. These results demonstrate that the proposed method GuFC has shown a better performance than the others, becoming a candidate to the segmentation step in a computational tool for coronary arteries visualization in the future

Selon le rapport annuel de la Fédération Mondiale du Cœur de 2023, les maladies cardiovasculaires (MCV) représentaient près d'un tiers de tous les décès mondiaux en 2021. Comparativement aux pays à revenu élevé, plus de 80% des décès par MCV surviennent dans les pays à revenu faible et intermédiaire. La répartition inéquitable des ressources de diagnostic et de traitement des MCV demeure toujours non résolue. Face à ce défi, les dispositifs abordables d'échographie de point de soins (POCUS) ont un potentiel significatif pour améliorer le diagnostic des MCV. Avec l'aide de l'intelligence artificielle (IA), le POCUS permet aux non-experts de contribuer, améliorant ainsi largement l'accès aux soins, en particulier dans les régions moins desservies.L'objectif de cette thèse est de développer des algorithmes robustes et automatiques pour analyser la fonction cardiaque à l'aide de dispositifs POCUS, en mettant l'accent sur l'échocardiographie et l'électrocardiogramme. Notre premier objectif est d'obtenir des caractéristiques cardiaques explicables à partir de chaque modalité individuelle. Notre deuxième objectif est d'explorer une approche multimodale en combinant les données d'échocardiographie et d'électrocardiogramme.Nous commençons par présenter deux nouvelles structures d'apprentissage profond (DL) pour la segmentation de l'échocardiographie et l'estimation du mouvement. En incorporant des connaissance a priori de forme et de mouvement dans les modèles DL, nous démontrons, grâce à des expériences approfondies, que de tels a priori contribuent à améliorer la précision et la généralisation sur différentes séries de données non vues. De plus, nous sommes en mesure d'extraire la fraction d'éjection du ventricule gauche (FEVG), la déformation longitudinale globale (GLS) et d'autres indices utiles pour la détection de l'infarctus du myocarde (IM).Ensuite, nous proposons un modèle DL explicatif pour la décomposition non supervisée de l'électrocardiogramme. Ce modèle peut extraire des informations explicables liées aux différentes sous-ondes de l'ECG sans annotation manuelle. Nous appliquons ensuite ces paramètres à un classificateur linéaire pour la détection de l'infarctus du myocarde, qui montre une bonne généralisation sur différentes séries de données.Enfin, nous combinons les données des deux modalités pour une classification multimodale fiable. Notre approche utilise une fusion au niveau de la décision intégrant de l'incertitude, permettant l'entraînement avec des données multimodales non appariées. Nous évaluons ensuite le modèle entraîné à l'aide de données multimodales appariées, mettant en évidence le potentiel de la détection multimodale de l'IM surpassant celle d'une seule modalité.Dans l'ensemble, nos algorithmes proposés robustes et généralisables pour l'analyse de l'échocardiographie et de l'ECG démontrent un potentiel significatif pour l'analyse de la fonction cardiaque portable. Nous anticipons que notre cadre pourrait être davantage validé à l'aide de dispositifs portables du monde réel
According to the 2023 annual report of the World Heart Federation, cardiovascular diseases (CVD) accounted for nearly one third of all global deaths in 2021. Compared to high-income countries, more than 80% of CVD deaths occurred in low and middle-income countries. The inequitable distribution of CVD diagnosis and treatment resources still remains unresolved. In the face of this challenge, affordable point-of-care ultrasound (POCUS) devices demonstrate significant potential to improve the diagnosis of CVDs. Furthermore, by taking advantage of artificial intelligence (AI)-based tools, POCUS enables non-experts to help, thus largely improving the access to care, especially in less-served regions.The objective of this thesis is to develop robust and automatic algorithms to analyse cardiac function for POCUS devices, with a focus on echocardiography (ECHO) and electrocardiogram (ECG). Our first goal is to obtain explainable cardiac features from each single modality respectively. Our second goal is to explore a multi-modal approach by combining ECHO and ECG data.We start by presenting two novel deep learning (DL) frameworks for echocardiography segmentation and motion estimation tasks, respectively. By incorporating shape prior and motion prior into DL models, we demonstrate through extensive experiments that such prior can help improve the accuracy and generalises well on different unseen datasets. Furthermore, we are able to extract left ventricle ejection fraction (LVEF), global longitudinal strain (GLS) and other useful indices for myocardial infarction (MI) detection.Next, we propose an explainable DL model for unsupervised electrocardiogram decomposition. This model can extract interpretable information related to different ECG subwaves without manual annotation. We further apply those parameters to a linear classifier for myocardial infarction detection, which showed good generalisation across different datasets.Finally, we combine data from both modalities together for trustworthy multi-modal classification. Our approach employs decision-level fusion with uncertainty, allowing training with unpaired multi-modal data. We further evaluate the trained model using paired multi-modal data, showcasing the potential of multi-modal MI detection to surpass that from a single modality.Overall, our proposed robust and generalisable algorithms for ECHO and ECG analysis demonstrate significant potential for portable cardiac function analysis. We anticipate that our novel framework could be further validated using real-world portable devices. We envision that such advanced integrative tools may significantly contribute towards better identification of CVD patients

The aim of this work is to propose an algorithm to segment and track the myocardium using the level-set formalism. The myocardium is first approximated by a geometric model (hyperquadrics) which allows to handle asymetric shapes such as the myocardium while avoiding a learning step. This representation is then embedded into the level-set formalism as a shape prior for the joint segmentation of the endocardial and epicardial borders. This shape prior term is coupled with a local data attachment term and a thickness term that prevents both contours from merging. The algorithm is validated on a dataset of 80 images at end diastolic and end systolic phase with manual references from 3 cardiologists. In a second step, we propose to segment whole sequences using motion information. To this end, we apply a level conservation constraint on the implicit function associated to the level-set and express this contraint as an energy term in a variational framework. This energy is then added to the previously described algorithm in order to constrain the temporal evolution of the contour. Finally the algorithm is validated on 20 echocardiographic sequences with manual references of 2 experts (corresponding to approximately 1200 images).

La thèse porte sur la segmentation spatio-temporelle et interactive d'images cardiaquesdynamiques. Elle s'inscrit dans le projet ANR 3DSTRAIN du programme"Technologies pour la Santé et l'Autonomie" qui a pour objectif d'estimer de façoncomplète, dense et sur plusieurs modalités d'imagerie 3D+t (telles que l'imageriepar résonance magnétique (IRM), la tomographie par émission monophotonique(TEMP) et l'échocardiographie) l'indice de déformation du muscle cardiaque : lestrain. L'estimation du strain nécessite une étape de segmentation qui doit être laplus précise possible pour fournir une bonne évaluation de cet indice. Nos travauxse sont orientés sur deux axes principaux : (1) le développement d'un modèle desegmentation conforme à la morphologie du muscle cardiaque et (2) la possibilitéde corriger interactivement et intuitivement le résultat de la segmentation obtenuegrâce à ce modèle
This thesis focuses on the spatio-temporal and interactive segmentation of dynamiccardiac images. It is a part of the ANR 3DSTRAIN project of program "Technologiesfor Health and Autonomy" which aims to estimate full, dense and on several3D+t imaging modalities (such as Magnetic Resonance Imaging (MRI), Single PhotonEmission Computed Tomography (SPECT) and echocardiography) the indexof deformation of the heart muscle : the strain. The strain estimation requires asegmentation step which must be as precise as possible to provide a good estimationof this index. Our work was focused on two main areas : (1) the development of asegmentation model conforms to the shape of the heart muscle and (2) the abilityto interactively and intuitively correct the segmentation's result obtained with thismodel

X-ray fluoroscopy and echocardiography imaging (ultrasound, US) are two imaging modalities that are widely used in cardiac catheterization. For these modalities, a fast, accurate and stable algorithm for the detection and tracking of catheters is required to allow clinicians to observe the catheter location in real-time. Currently X-ray fluoroscopy is routinely used as the standard modality in catheter ablation interventions. However, it lacks the ability to visualize soft tissue and uses harmful radiation. US does not have these limitations but often contains acoustic artifacts and has a small field of view. These make the detection and tracking of the catheter in US very challenging. The first contribution in this thesis is a framework which combines Kalman filter and discrete optimization for multiple catheter segmentation and tracking in X-ray images. Kalman filter is used to identify the whole catheter from a single point detected on the catheter in the first frame of a sequence of x-ray images. An energy-based formulation is developed that can be used to track the catheters in the following frames. We also propose a discrete optimization for minimizing the energy function in each frame of the X-ray image sequence. Our approach is robust to tangential motion of the catheter and combines the tubular and salient feature measurements into a single robust and efficient framework. The second contribution is an algorithm for catheter extraction in 3D ultrasound images based on (a) the registration between the X-ray and ultrasound images and (b) the segmentation of the catheter in X-ray images. The search space for the catheter extraction in the ultrasound images is constrained to lie on or close to a curved surface in the ultrasound volume. The curved surface corresponds to the back-projection of the extracted catheter from the X-ray image to the ultrasound volume. Blob-like features are detected in the US images and organized in a graphical model. The extracted catheter is modelled as the optimal path in this graphical model. Both contributions allow the use of ultrasound imaging for the improved visualization of soft tissue. However, X-ray imaging is still required for each ultrasound frame and the amount of X-ray exposure has not been reduced. The final contribution in this thesis is a system that can track the catheter in ultrasound volumes automatically without the need for X-ray imaging during the tracking. Instead X-ray imaging is only required for the system initialization and for recovery from tracking failures. This allows a significant reduction in the amount of X-ray exposure for patient and clinicians.

This thesis describes design and implementation of method, which determines beginning of heart beats in echocardiographic record. Design of this method is built around pyramidal Lucas-Kanade algorithm and fast Fourier transform. This method is implemented in C++ language with OpenCV and FFTW libraries. Analysis of the implementation has shown, that this method is sensitive to anomalies in echocardiographic record. This method is developed as a part of the project for an analysis of echocardiographic records for st. Anna hospital at Brno.

Speckle noise appears when coherent illumination is employed, as for example Laser, Synthetic Aperture Radar (SAR), Sonar, Magnetic Resonance, X-ray and Ultrasound imagery. Backscattered echoes from the randomly distributed scatterers in the microscopic structure of the medium are the origin of speckle phenomenon, which characterizes coherent imaging with a granular appearance. It can be shown that speckle noise is of multiplicative nature, strongly correlated and more importantly, with non-Gaussian statistics. These characteristics differ greatly from the traditional assumption of white additive Gaussian noise, often taken in image segmentation, filtering, and in general, image processing; which leads to reduction of the methods effectiveness for final image information extraction; therefore, this kind of noise severely impairs human and machine ability to image interpretation. Statistical modeling is of particular relevance when dealing with speckled data in order to obtain efficient image processing algorithms; but, additionally, clinical ultrasound imaging systems employ nonlinear signal processing to reduce the dynamic range of the input echo signal to match the smaller dynamic range of the display device and to emphasize objects with weak backscatter. This reduction in dynamic range is normally achieved through a logarithmic amplifier i.e. logarithmic compression, which selectively compresses large input signals. This kind of nonlinear compression totally changes the statistics of the input envelope signal; and, a closed form expression for the density function of the logarithmic transformed data is usually hard to derive. This thesis is concerned with the statistical distributions of the Log-compressed amplitude signal in coherent imagery, and its main objective is to develop a general statistical model for log-compressed ultrasound B-scan images. The developed model is adapted, making the pertinent physical analogies, from the multiplicative model in Synthetic Aperture Radar (SAR) context. It is shown that the proposed model can successfully describe log-compressed data generated from different models proposed in the specialized ultrasound image processing literature. Also, the model is successfully applied to model in-vivo echo-cardiographic (ultrasound) B-scan images. Necessary theorems are established to account for a rigorous mathematical proof of the validity and generality of the model. Additionally, a physical interpretation of the parameters is given, and the connections between the generalized central limit theorems, the multiplicative model and the compound representations approaches for the different models proposed up-to-date, are established. It is shown that the log-amplifier parameters are included as model parameters and all the model parameters are estimated using moments and maximum likelihood methods. Finally, three applications are developed: speckle noise identification and filtering; segmentation of in vivo echo-cardiographic (ultrasound) B-scan images and a novel approach for heart ejection fraction evaluation
El ruido Speckle aparece cuando se utilizan sistemas de iluminación coherente, como por ejemplo Láser, Radar de Apertura Sintética (SAR), Sonar, Resonancia Magnética, rayos X y ultrasonidos. Los ecos dispersados por los centros dispersores distribuidos al azar en la estructura microscópica del medio son el origen de este fenómeno, que caracteriza las imágenes coherentes con un aspecto granular. Se puede demostrar que el ruido Speckle es de carácter multiplicativo, fuertemente correlacionados y lo más importante, con estadística no Gaussiana. Estas características son muy diferentes de la suposición tradicional de ruido aditivo gaussiano blanco, a menudo asumida en la segmentación de imágenes, filtrado, y en general, en el procesamiento de imágenes; lo cual se traduce en la reducción de la eficacia de los métodos para la extracción de información de la imagen final. La modelización estadística es de particular relevancia cuando se trata con datos Speckle, a fin de obtener algoritmos de procesamiento de imágenes eficientes. Además, el procesamiento no lineal de señales empleado en sistemas clínicos de imágenes por ultrasonido para reducir el rango dinámico de la señal de eco de entrada de manera que coincida con el rango dinámico más pequeño del dispositivo de visualización y resaltar así los objetos con dispersión más débil, modifica radicalmente la estadística de los datos. Esta reducción en el rango dinámico se logra normalmente a través de un amplificador logarítmico es decir, la compresión logarítmica, que comprime selectivamente las señales de entrada y una forma analítica para la expresión de la función de densidad de los datos transformados logarítmicamente es por lo general difícil de derivar. Esta tesis se centra en las distribuciones estadísticas de la amplitud de la señal comprimida logarítmicamente en las imágenes coherentes, y su principal objetivo es el desarrollo de un modelo estadístico general para las imágenes por ultrasonido comprimidas logarítmicamente en modo-B. El modelo desarrollado se adaptó, realizando las analogías físicas relevantes, del modelo multiplicativo en radares de apertura sintética (SAR). El Modelo propuesto puede describir correctamente los datos comprimidos logarítmicamente a partir datos generados con los diferentes modelos propuestos en la literatura especializada en procesamiento de imágenes por ultrasonido. Además, el modelo se aplica con éxito para modelar ecocardiografías en vivo. Se enuncian y demuestran los teoremas necesarios para dar cuenta de una demostración matemática rigurosa de la validez y generalidad del modelo. Además, se da una interpretación física de los parámetros y se establecen las conexiones entre el teorema central del límite generalizado, el modelo multiplicativo y la composición de distribuciones para los diferentes modelos propuestos hasta a la fecha. Se demuestra además que los parámetros del amplificador logarítmico se incluyen dentro de los parámetros del modelo y se estiman usando los métodos estándar de momentos y máxima verosimilitud. Por último, tres aplicaciones se desarrollan: filtrado de ruido Speckle, segmentación de ecocardiografías y un nuevo enfoque para la evaluación de la fracción de eyección cardiaca.

博士
國立交通大學
資訊科學系所
92
In this dissertation, we propose two region-based segmentation algorithms and two edge-based segmentation algorithms for echocardiographic images. The first proposed algorithm of region-based segmentation scheme is fuzzy Hopfield neural network with fixed weight approach. This approach incorporated the global gray-level information and local gray-level information to construct a fuzzy Hopfield neural network. When the network converges to a stable state, the segmentation result will be obtained. A new approach using alpha-shape points is another proposed algorithm of region-based category. The region of interest corresponds to one of the clusters under a properly selected alpha. We identify the heart chamber in the ultrasound image by comparing the similarity between the alpha-connected components against the heart chamber obtained from the AQ image. The first proposed algorithm of edge-based segmentation scheme is finding the shortest path in directed graph. We circularly spread the image first and then map it to a directed graph. To avoid the local minimum trapping, dynamic programming approach is used for finding the shortest path. The other proposed approach for edge-based segmentation algorithm is suitable for non-circular like boundary. We incorporated an alpha-contour approach based on alpha-shape technique to construct the search space and then map it to a directed graph. The dynamic programming technique also used for finding the shortest path. In additional, we also propose a new approach for extracting mitral annular lines for echocardiographic images. A nearly automatic method for calculating the mitral annular lines from a 2D+1D precordial echocardiogram four-chamber view was presented. The proposed method needs only a physician to provide a point in the left ventricular chamber. The average error was 3% which is clinically acceptable. The proposed method saves much clinician time, allowing a shift from machine to patient care.

碩士
國立中正大學
資訊工程所
95
Cardiovascular disease is generally the number one cause of death in the Western world. Segmentation and tracking of cardiac structures are advanced techniques used to assist physicians in various states of treatment of cardiovascular diseases. Due to the advantages of convenience, non-invasive technique, and real-time operation, the three-dimensional (3-D) echocardiography has become a valuable diagnostic imaging modality for patients with cardiovascular diseases. A complete cardiac cycle includes two significant phases: systole and diastole that the volumes of both will be calculated to evaluate the ejection fraction (EF). The EF is one of the crucial indicators for the evaluation of cardiac pump functions. Therefore, the extraction of the left ventricle (LV) from echocardiographic images and particularly evaluates the ejection fraction in cardiac cycle, is a very useful step for clinical diagnosis. This paper presents an automatic system to extract the contour of the LV and evaluate the EF in cardiac cycle. The segmentation method uses a fuzzy reasoning and relaxation techniques to extract the endocardial boundaries from 3-D echocardiographic images. This method was validated against the physician-defined segmentation and demonstrated with acceptable accuracy. In this paper, totally 37 3-D cardiac ultrasound images are validated that the average accuracy rate is 90.29%. Also, the EF results between the proposed method segmentation and physician-defined segmentation have highly positive correlation. These two results present that this automatic system can extract the LV contour and obtain the EF value effectively.

Elaborazione di sequenze di immagini con metodi variazionali.
Analysis of left ventricle echocardiographic movies by vari-ational methods: a dedicated software for ECG and ECHO syn-chronizations during cardiac cycles. The image sequences elaboration, which represent our main topic, concerns the identi cation of contours of an object for segmentation and study of its movement over time. A newer dynamic approach to the well-known static variational method for the time-series medical echocardiographic images is presented, and a graphic software applications for synchronization between cardiac movement and electric signals (ECG) are developed. Many approaches have been proposed to process time-series of digital images, and it is di cult establish the most e ective one. Here we focus on PDE-based and variational methods. Standard snake models of closed curve evolution, pertaining to the Level-Set Method due to S. Osher and J. A. Sethian et al., have been implemented to characterize the internal ventricular area over time. We have applied standard nite di erence techniques for the approximation solution of the involved eikonal equations for front evolution. Well-known techniques of Gaussian regularization, such as heat equation, have highlighted the loss of de nition of ventricular edge. Thus we now present a numerical frame preprocessing technique, based on a variational model, that consists of functional minimization with a Mumford-Shah (M-S) time-dependent energy term, which is suitable to enhance ventricular edge and regularize initial data for curve evolution. This one, unlike the standard convolution with a regularizing operator, allows to compute the bright-intensity gradient of the image for velocity term of the curve evolution. We show the formulation of the standard model problem for image segmentations by curve evolution and a newer approach of pre-treating model by functional minimization. The applicability of the resolution method is discussed and an open mathematical problem for a new functional involving time-dependence are presented. Finally we describes the digital format of movies, the adaptation of single frames, the step which characterize the elaboration protocol, the quality of the results obtained with the preprocessing on a test problem. We additionaly describe the choice and implementation of reduced numerical methods in the real case and their computational cost. We compare results obtained by various preprocessing methods on identi ed areas, the curve evolution beyond the edge, ventricular area his trends over time and the evaluated Ejection Fraction. The implemented graphic application is presented, as well as the principal use of the signal-synchronizing software.

Tese de Doutoramento em Engenharia Biomédica
Aortic stenosis (AS), a condition in which the aortic valve (AV) calcifies and obstructs the blood flow from the left ventricle to the aorta, is the most common acquired valvular heart disease in the elderly population. Among the available therapies, transcatheter AV implantation (TAVI) has had a continuously growing worldwide indication for patients with symptomatic, severe AS. During this minimally invasive percutaneous intervention, a prosthetic valve is delivered inside a catheter and deployed under imaging guidance at the aortic annulus’ level, without removing the native damaged valve. Despite its high success rate, procedural complications are frequent, including moderate or severe paravalvular regurgitation (PVR), damage to neighbor structures, heart conduction disturbances requiring a permanent pacemaker implantation, among others. In spite of numerous clinical and technological advances presented since its introduction by all stakeholders involved, there is still a lack of consensus regarding the ideal clinical, imaging and technological setting to plan it, to perform it and to evaluate its outcomes. Notwithstanding, it is accepted that most of the aforesaid complications could be avoided, or at least their impact minimized, by an adequate preoperative planning combined with a precise intraoperative image-based guidance, and followed by an accurate postoperative quality assurance of the implanted prosthesis. With this in mind, this work aimed at developing novel medical imaging processing solutions to help throughout all TAVI stages, focusing particularly in potentiating the use of 3D transesophageal echocardiography (TEE) within its clinical workflow. Aiming at an adequate preoperative planning, a fully automatic pipeline for AV tract segmentation and prosthesis sizing from 3D-TEE images was developed and validated. This framework proposed, for the first time, the extension of the B-spline Explicit Active Surfaces (BEAS) framework towards a shapebased deformable model strategy, allowing to combine BEAS’ computational efficiency with the shape information provided by statistical shape models. The framework was shown to be feasible, accurate, robust and time efficient, allowing to extract accurate measurements without user-induced variability while significantly reducing the analysis time. This pipeline was then integrated in an in-house visualization and analysis software suite, combined with interactive tools, and validated in an extensive clinical database with both 3D-TEE and multidetector computer tomography (MDCT) images. Overall, the software-based prosthesis sizes were concordant with those extracted by MDCT- and TEE-based manual sizing strategies, ultimately demonstrating the software’s potential interest to facilitate the analysis of 3D-TEE images. Moreover, to allow a multimodality planning, this framework was then adapted for a fully automatic assessment of MDCT images, showing an excellent agreement with manually-derived measurements. A semi-automatic framework for 3D cardiac magnetic resonance images was also proposed. Aiming at further assisting the planning stage and help during the intraoperative scenario, a novel algorithm for efficient and robust image-based tracking in 3D-TEE images was proposed. Hereto, a novel localized anatomical affine optical flow tracking (AAOF) strategy was presented, showing an increased ability to capture an object’s local motion and deformation. Based on this algorithm, a two-step tracking method was proposed for fast, automatic AV tract tracking in 3D-TEE images, which may be used to extract dynamic AV measurements at the TAVI planning or augment these images with relevant medical annotations during intraoperative guidance. The methodology was quantitatively validated for the former, demonstrating to be feasible, accurate and robust. Furthermore, given the vast clinical and research interest in image tracking solutions, several variants of the proposed object-based image tracking method were combined in a novel user-friendly software package – the Medical Image Tracking Toolbox (MITT). This toolbox was designed to ease the customization of image tracking solutions in the medical field, allowing to track multiple types of objects in both 2D and 3D image sequences. Illustrative examples from the cardiology field were provided, demonstrating the versatility, simplicity and time efficiency of MITT. Finally, aiming at helping the development of postoperative quality assurance tools, synthetic 3D color Doppler ultrasound (US) images that mimic cases of post-implantation PVR in TAVI patients were generated using an US simulation-based pipeline. These simulated volumes can mimic distinct levels of PVR severity, and may be used to validate novel methodologies for PVR quantification or to study clinicallyused color Doppler parameters for PVR grading. A preliminary clinical study focusing on the latter was presented, shedding light on important issues that must be taken into account during PVR quantification. Overall, this work contributed with a vast set of intertwined algorithms, which may aid physicians at the pre-, intra- and postoperative stages of TAVI, and ultimately ease and improve its clinical workflow and thus benefit both the patients and the physicians.
A estenose aórtica, doença na qual a válvula aórtica calcifica e obstrui o fluxo de sangue do ventrículo esquerdo para a aorta, é a valvulopatia adquirida mais comum na população idosa. Entre as terapias disponíveis, a implantação valvular aórtica percutânea (IVAP) tem tido uma aceitação crescente a nível mundial para pacientes sintomáticos com estenose aórtica grave. Nesta intervenção minimamente invasiva, uma válvula prostética é posicionada, com recurso a imagiologia médica, ao nível do anel aórtico, sem remover a válvula nativa danificada. Apesar da alta taxa de sucesso, complicações no decorrer desta intervenção são frequentes, e incluem regurgitação paravalvular (RPV) moderada ou severa, dano nas estruturas vizinhas, distúrbios da condução elétrica cardíaca com necessidade de implantação de pacemaker permanente, entre outras. Embora múltiplos avanços clínicos e tecnológicos tenham sido propostos desde a sua introdução, ainda existe uma falta de consenso quanto ao cenário clínico, imagiológico e tecnológico ideal para planear, executar ou avaliar os resultados desta intervenção. Não obstante, é aceite que muitas das complicações acima mencionadas podem ser evitadas, ou pelo menos o seu impacto minimizado, por um planeamento pré-operatório adequado, combinado com uma implantação intraoperatória guiada por imagem precisa, e seguido por uma avaliação pós-operatória do sucesso obtido na colocação da válvula prostética. Deste modo, este trabalho visou o desenvolvimento de novas soluções de processamento de imagem médica para ajudar nos vários estágios da IVAP, focando em potenciar particularmente o uso do ecocardiograma transesofágico (ETE) tridimensional em todas as fases desta intervenção. No sentido de melhorar o planeamento pré-operatório, foi desenvolvido e validado um algoritmo automático de segmentação do trato aórtico e de dimensionamento da válvula prostética a partir de imagens de 3D ETE. Este algoritmo propõe a extensão da framework “B-spline Explicit Active Surfaces” (BEAS) para uma técnica de modelos deformáveis baseados na forma. O algoritmo automático mostrou ser viável, preciso, robusto e computacionalmente eficiente, permitindo a extração de medidas precisas sem, contudo, depender do utilizador e reduzindo significativamente o tempo de análise. Este algoritmo foi ainda integrado num software de visualização e análise de imagens, combinado com ferramentas interativas, e validado numa base de dados clínica extensa com imagens de 3D ETE e de tomografia axial computadorizada (TAC). O dimensionamento automatizado da prótese mostrou ser concordante com as medidas extraídas por estratégias manuais de dimensionamento em imagens de 3D ETE e TAC. Além disso, de modo a permitir um planeamento baseado em múltiplas modalidades de imagem, o algoritmo proposto foi adaptado para um dimensionamento automático em imagens de TAC, mostrando resultados excelentes quando comparados com medidas extraídas manualmente. Mais ainda, foi proposto um algoritmo semiautomático adaptado para imagens de ressonância magnética cardíaca 3D. Com o objetivo de ajudar na intervenção, mas também auxiliar o planeamento, foi proposto um novo algoritmo para tracking de imagens de 3D ETE, denominado “localized anatomical affine optical flow”. Esta estratégia provou ser capaz de capturar localmente os movimentos e deformações do objeto em estudo. Com base neste algoritmo, foi proposta uma nova metodologia rápida e automática para tracking da parede do trato aórtico em imagens de 3D ETE, a qual pode ser usada para extrair medidas dinâmicas da válvula aórtica durante a fase de planeamento ou para sobrepor anotações médicas em imagens intraoperatórias com vista a simplificar a implantação da válvula prostética. A metodologia foi validada para a extração dinâmica de medidas, demonstrando ser viável, precisa e robusta. Além disso, dado o vasto interesse clínico e de investigação em soluções de tracking de imagens, várias variantes do algoritmo proposto foram combinadas numa toolbox, denominada “Medical Image Tracking Toolbox”. Esta toolbox foi projetada para facilitar a personalização de soluções de tracking de imagens médicas, permitindo efetuar o tracking de múltiplos tipos de objetos em imagens dinâmicas 2D e 3D. A versatilidade, simplicidade e eficiência computacional desta toolbox foi demonstrada com recurso a múltiplos exemplos de aplicações em cardiologia. Por fim, e com o intuito de melhorar o atual processo de avaliação pós-operatória na identificação de possíveis complicações, foi demonstrado o uso de um simulador de imagens de ultrassom para gerar volumes de ultrassonografia com Doppler colorido que simulam casos de RPV pós-operatória, com níveis de severidade distintos, em pacientes de IVAP. Estas imagens sintéticas podem ser usadas para validação de novas metodologias para quantificação da RPV ou para estudar parâmetros clínicos extraídos a partir deste tipo de imagens e usados para classificar a severidade desta. Um estudo clínico preliminar de alguns destes parâmetros foi apresentado, evidenciando fatores importantes a serem tidos em conta para a quantificação precisa da regurgitação paravalvular pós-IVAP. Em suma, este trabalho contribuiu com um vasto conjunto de algoritmos interligados, os quais visam auxiliar os intervencionistas nos estágios pré-, intra- e pós-operatórios da IVAP e, assim, facilitar e melhorar o procedimento clínico em geral e beneficiar tanto os pacientes como os próprios médicos.
The present work was possible thanks to the financial support provided by FCT - Foundation for Science and Technology, and the European Social Fund, through the Programa Operacional Capital Humano (POCH), in the scope of the PhD grant SFRH/BD/93443/2013; by funds from the European Regional Development Fund (FEDER) through the Operational Programme Competitiveness Factors (COMPETE) and by National Funds through FCT under the projects POCI-01-0145-FEDER-007038 and UID/CEC/00319/2013; and by the project NORTE-01-0145-FEDER-000013, supported by the Northern Portugal Regional Operational Programme (NORTE 2020), under the PORTUGAL 2020 Partnership Agreement, through FEDER.