[pt] Na indústria do petróleo, problemas como dano mecânico causam redução da porosidade e permeabilidade de uma formação rochosa, reduzindo a produtividade e injetividade de poços de sistemas de produção de óleo e gás. Na perfuração do poço há alteração do estado de tensões no seu entorno, causando uma deformação na rocha que pode induzir a uma perda significativa da permeabilidade. Nesta dissertação foi realizado um estudo da influência do dano mecânico na porosidade de rochas do tipo arenito. Para isso, utilizou-se a técnica não-destrutiva de microtomografia de raios-x, que permite a visualização da estrutura interna de materiais, acoplada a uma célula desenvolvida para aplicação, in situ, de tensão hidrostática. Uma amostra de arenito como 8 mm de diâmetro foi tomografada em 3 condições: sem carregamento, após a aplicação de tensão hidrostática de 3300 psi e após o descarregamento. A célula permitiu que as variações de carga fossem realizadas sem retirar a amostra do tomógrafo, permitindo uma comparação quantitativa entre as imagens 3D. Nas 3 condições foram obtidos dados como porosidade total, variação da área porosa em cada camada, volume e forma dos poros. / [en] In the oil industry, problems such as mechanical damage reduce the porosity and permeability of a rock formation, reducing the productivity and injectivity of wells in oil and gas production systems. During the well drilling there is a change in the state of the stress in its surroundings, causing a deformation in the rock that can induce a significant loss of permeability. In this dissertation, it was carried out a study regarding the influence of mechanical damage on the porosity of sandstone rocks. In order to do this, the non-destructive technique of x-ray microtomography was used, which allows the visualization of the materials internal structure, coupled to a cell developed for in situ application of hydrostatic stress. A sandstone sample of 8 mm in diameter was scanned under 3 conditions: without load, after application of 3300 psi hydrostatic stress and after unloading. The cell allowed the load variations to be performed without removing the sample from the tomograph, allowing a quantitative comparison between the 3D images in the 3 conditions. Data such as total porosity, variation of the porous area in each layer, volume and shape of the pores were obtained.

Computed tomography is one of the most evolving fields of modern radiology. The current CT applications permit among other things angiography, 3D image reconstructions, material decomposition and tissue characterization. CT is an important tool in the assessment of specific patient populations prior to an invasive or surgical procedure. The aim of this dissertation was to demonstrate the decisive role of CT with 3D-image reconstructions in haemodialysis patients scheduled to undergo fistulography, in patients undergoing surgical breast reconstructions with a perforator flap and in patients with complicated renal calculi scheduled to undergo percutaneous nephrolithotomy. CT Angiography with 3D image reconstructions was performed in 31 patients with failing arteriovenous fistulas and grafts, illustrating the vascular anatomy in a comprehensive manner in 93.5% of the evaluated segments and demonstrating a sensitivity of 95% compared to fistulography. In 59 mastectomy patients scheduled to undergo reconstructive breast surgery with a deep inferior epigastric perforator flap, the preoperative planning with CT Angiography with 3D image reconstructions of the anterior abdominal wall providing details of its vascular supply, reduced surgery time significantly (p< 0.001) and resulted in fewer complications. Dual Energy CT Urography with advanced image reconstructions in 31 patients with complicated renal calculi scheduled to undergo PNL, resulted in a new method of material characterisation (depicting renal calculi within excreted contrast) and in the possibility of reducing radiation dose by 28% by omitting the nonenhanced scanning phase. Detailed analysis of the changes renal calculi undergo when virtually reconstructed was performed and a comparison of renal calculi number, volume, height and attenuation between virtual nonenhanced and true nonenhanced images was undertaken. All parameters were significantly underestimated in the virtual nonenhanced images. CT with 3D-reconstructions is more than just “flashy images”. It is crucial in preoperative planning, optimizes various procedures and can reduce radiation dose.

Today modern industry systems are almost fully automated. The high requirements regarding speed, flexibility, precision and reliability makes it in some cases very difficult to create. One of the most willingly researched solution to solve many processes without human influence is bin-picking. Bin picking is a very complex process which integrates devices such as: robotic grasping arm, vision system, collision avoidance algorithms and many others. This paper describes the creation of a vision system - the most important part of the whole bin-picking system. Authors propose a model-based solution for estimating a best pick-up candidate position and orientation. In this method database is created from 3D CAD model, compared with processed image from the 3D scanner. Paper widely describes database creation from 3D STL model, Sick IVP 3D scanner configuration and creation of the comparing algorithm based on autocorrelation function and morphological operators. The results shows that proposed solution is universal, time efficient, robust and gives opportunities for further work. / +4915782529118

This thesis deals with possibilities of automatic segmentation of biomedical images. For the 3D image segmentation, a deep learning method has been proposed. In the work problems of network design, memory optimization method and subsequent composition of the resulting image are solved. The uniqueness of the method lies in 3D image processing on a GPU in combination with augmentation of training data and preservation of the output size with the original image. This is achieved by dividing the image into smaller parts with the overlay and then folding to the original size. The functionality of the method is verified on the segmentation of human brain tissue on magnetic resonance imaging, where it overcomes human accuracy when compared a specialist vs. specialist, and cell segmentation on a slices of the Drosophila brain from an electron microscope, where published results from the impacted paper are overcome.

The topic of this thesis is the design of dental X-ray. User problems were identified based on the design and technical analysis for which solutions are presented in the work. The innovative shape and design solution shows a new approach to dental X-rays. The main benefit of the design is the solution to real ergonomic problems that can occur during the interaction of the operator or patient and the device. The design corresponds to current trends in the evolving design field of medical design.

Based on the generating matrix to the Tribonacci sequence, the Tribonacci cat map is a discrete chaotic dynamical system, similar to Arnold's discrete cat map, but on three dimensional space. In this thesis, this new mapping is introduced and the properties of its matrix are presented. The main results of the investigation prove how the size of the domain of the map affects its period and explore the orbit lengths of non-trivial points. Different upper bounds to the map are studied and proved, and a conjecture based on numerical calculations is proposed. The Tribonacci cat map is used for applications such as 3D image encryption and colour encryption. In the latter case, the results provided by the mapping are compared to those from a generalised form of the map.

No description available.

Heart disease is the major cause of death in the developed world. Due to its fast, portable, low-cost and harmless way of imaging the heart, echocardiography has become the most frequent tool for diagnosis of cardiac function in clinical routine. However, visual assessment of heart function from echocardiography is challenging, highly operatordependant and is subject to intra- and inter observer errors. Therefore, development of automated methods for echocardiography analysis is important towards accurate assessment of cardiac function. In this thesis we develop new ways to model echocardiography data using Bayesian machine learning methods and concern three problems: (i) wall motion analysis in 2D stress echocardiography, (ii) segmentation of the myocardium in 3D echocardiography, and (iii) standard views extraction from 3D echocardiography. Firstly, we propose and compare four discriminative methods for feature extraction and wall motion classification of 2D stress echocardiography (images of the heart taken at rest and after exercise or pharmalogical stress). The four methods are based on (i) Support Vector Machines, (ii) Relevance Vector Machines, (iii) Lasso algorithm and Regularised Least Squares, (iv) Elastic Net regularisation and Regularised Least Squares. Although all the methods are shown to have superior performance to the state-of-the-art, one conclusion is that good segmentation of the myocardium in echocardiography is key for accurate assessment of cardiac wall motion. We investigate the application of one of the most promising current machine learning techniques, called Decision Random Forests, to segment the myocardium from 3D echocardiograms. We demonstrate that more reliable and ultrasound specific descriptors are needed in order to achieve the best results. Specifically, we introduce two sets of new features to improve the segmentation results: (i) LoCo and GloCo features with a local and a global shape constraint on coupled endoand epicardial boundaries, and (ii) FA features, which use the Feature Asymmetry measure to highlight step-like edges in echocardiographic images. We also reinforce the traditional features such as Haar and Rectangular features by aligning 3D echocardiograms. For that we develop a new registration technique, which is based on aligning centre lines of the left ventricles. We show that with alignment performance is boosted by approximately 15%. Finally, a novel approach to detect planes in 3D images using regression voting is proposed. To the best of our knowledge we are the first to use a one-step regression approach for the task of plane detection in 3D images. We investigate the application to standard views extraction from 3D echocardiography to facilitate efficient clinical inspection of cardiac abnormalities and diseases. We further develop a new method, called the Class- Specific Regression Forest, where class label information is incorporating into the training phase to reinforce the learning from semantically relevant to the problem classes. During testing the votes from irrelevant classes are excluded from voting to maximise the confidence of output predictors. We demonstrate that the Class-Specific Regression Random Forest outperforms the classic Regression Random Forest and produces results comparable to the manual annotations.

Dans l'objectif de développer des nouveaux capteurs d'image 3D pour des applications émergeantes, nous avons étudié un pixel de mesure de distance de type « temps de vol ». Nous avons proposé une nouvelle architecture de pixel basée sur la méthode « Continuous-Wave modulation » à trois échantillons par pixel. Cette méthode repose sur la mesure d'un déphasage entre la source lumineuse modulée en amplitude envoyée (source proche infrarouge) et le signal réfléchi par la scène à capturer. Le pixel de dimensions 6,2μm x 6,2μm intègre une photodiode pincée, trois chemins de transfert de charges pour l'échantillonnage successif du signal modulé reçu, et d'un quatrième chemin pour évacuer les charges excédentaires. Les différents chemins de transfert sont constitués d'une grille de transfert de charges de la photodiode vers une mémoire de stockage à canal enterré pour améliorer le rendement et la vitesse de transfert de charges; d'une mémoire à stockage en volume à base de tranchées capacitives profondes afin d'augmenter la dynamique; d'un substrat dont l'épaisseur et le profil de dopage ont été optimisés afin de collecter efficacement les charges photogénérées et ainsi augmenter les performances de démodulation. Un véhicule de test constitué d'une matrice de résolution de 464x197 pixels (QVGA) a été fabriqué, différentes variantes de pixels et différents essais technologiques ont été étudiées et analysées. La fonctionnalité du pixel a été vérifiée pour des fréquences de démodulation de 20MHz à 165MHz, utilisant une source laser de longueur d'onde 850nm ou 950nm. Une première image de profondeur acquise utilisant une matrice de test est une validation du pixel proposé / In order to develop new 3D image sensors for emerging applications, we studied “time of flight” pixel for distance measurement. We have proposed a new pixel architecture based on the "Continuous-Wave Modulation" method with three samples per pixel. This method is based on the measurement of a phase shift between the transmitted amplitude modulated light source (near-infrared source) and the signal reflected by the scene to be captured. The pixel of dimensions 6.2 μm x 6.2 μm integrates a pinned photodiode, three charge transfer paths for successive sampling of the received modulated signal, and a fourth path for anti-blooming purpose. The different paths are controlled by a buried-channel transfer gate for charges transfer from the photodiode to memory in order to improve the efficiency and speed of the charge transfer; A fully depleted memory based on capacitive deep trenches is used to increase the memory storage capacitance; thickness and doping profile of the substrate have been optimized to efficiently collect photogenerated and increase demodulation performance. The designed 464x197-pixel (QVGA) test chip has been fabricated, different pixel variants and different technology trials have been studied and analyzed. Pixel functionality has been verified for demodulation frequencies from 20 to 165MHz, using a laser source of wavelength 850nm or 950nm. A first acquired depth image using the test chip made is a validation of the proposed pixel

Dans l’optique de refroidir les parois des chambres de combustion aéronautiques le plus efficacement possible, un intérêt particulier est aujourd’hui porté à la technologie de refroidissement par transpiration. L’air de refroidissement s’écoule au travers d’une paroi poreuse dans laquelle une grande quantité de chaleur est échangée par convection. L’éjection de l’air profite ensuite de la distribution des pores pour former une couche limite protectrice relativement homogène.Les matériaux métalliques obtenus à partir de poudres partiellement frittées sont de bons candidats pour former ces parois poreuses. Ce travail se focalise sur les échanges internes et consiste à développer une méthodologie permettant de dégager les architectures partiellement frittées les plus adaptées à ce type d’application.L’écoulement et les échanges de chaleur lors du refroidissement par transpiration sont régis par quelques propriétés effectives des matériaux qui sont fonction de l’architecture : la conductivité thermique effective, le coefficient de transfert convectif volumique et les propriétés de perméabilité. A l’aide de travaux expérimentaux ou d’études numériques sur des échantillons numérisés par tomographie aux rayons X, des relations simples entre les propriétés effectives des matériaux partiellement frittés et leurs paramètres architecturaux sont tout d’abord développées. La porosité, la surface spécifique et le type de poudre utilisé sont retenus pour prédire les paramètres effectifs.Ces relations sont finalement intégrées dans un modèle de transfert de chaleur prédisant la performance d’une solution dans les conditions de fonctionnement du moteur. Une optimisation "multi-objectifs" et une analyse des designs optimaux permettent alors de mettre en valeur quelques architectures montrant un fort potentiel pour des applications de refroidissement par transpiration. Des matériaux peu poreux formés à partir de larges poudres irrégulières semblent assurer le meilleur compromis entre tous les critères pris en compte. / In order to cool aero-engine combustion chambers as efficiently as possible, there is today a special interest given to transpiration cooling technology. The cooling air flows through a porous liner in which a large amount of heat can be exchanged by convection. The air injection could then take benefit of the pore distribution to form a more homogeneous protective boundary layer.Partially sintered metallic materials are potential candidates to form these porous liners. The present work focuses on internal heat transfers. It aims to develop a methodology capable of highlighting the most adapted partially sintered architectures to this kind of application.During transpiration cooling, flows and heat transfers are governed by some effective material properties which depends on the porous architecture: the effective solid phase thermal conductivity, the volumetric heat transfer coefficient and the permeability properties. Thanks to experimental works and numerical studies on samples digitized by X-ray tomography, simple relationships are first developed between the effective material properties of partially sintered materials and their architectural parameters. The porosity, the specific surface area and the powder type are selected to predict the effective properties.These relationships are finally integrated into a heat transfer model predicting the thermal performance of a design at working engine conditions. A multi-objective optimization and an analysis of the optimal designs highlight some architectures as being potentially interesting for transpiration cooling. Materials with a low porosity and made of large irregular powders seem to ensure the best trade-off among the different criteria taken into consideration.