A resistência tubular renal tem sido estudada a fim de se ampliar a compreensão da fisiopatologia da Insuficiência renal aguda (IRA). A isquemia renal induz à resistência a um subseqüente insulto isquêmico sendo que os mecanismos de resistência parecem depender de alterações celulares. O p21 é um inibidor do ciclo celular, o qual pode ser induzido por radicais livres de oxigênio e parece ter um efeito protetor na IRA isquêmica. O objetivo deste estudo é avaliar o papel do p21 e do estresse oxidativo em modelo de resistência adquirida após episódio de IRA isquêmica, e em túbulos proximais isolados após isquemia. Ratos Wistar foram divididos em 3 grupos: grupo 1- sham, grupo 2- submetido a procedimento sham e após 2 dias submetido à isquemia de 45 min e grupo 3- submetido à isquemia de 45 min e após 2 dias submetido à segunda isquemia de 45 min. Os valores de uréia plasmática (114±60 vs. 136±44 mg/dL, n.s.), a creatinina sérica (0,86±0,2 vs. 0,98±0,1mg/dL, n.s.) e o clearance de creatinina (0,21±0,1vs. 0,24±0,1mL/min/100g, n.s.), avaliados 48 h após o segundo procedimento (Dia 4), foram semelhantes entre os grupos 2 e 3. O tempo de recuperação da IRA também foi semelhante entre os grupos 2 e 3. A histologia mostrou necrose tubular aguda aparentemente de grau semelhante entre os grupos 2 e 3. O infiltrado linfocitário foi semelhante entre os 3 grupos, entretanto houve aumento no infiltrado de macrófagos no grupo 3. Foi observado aumento na proliferação celular no grupo 2 e grupo 3, quando comparados ao grupo 1(125±28 cél./mm2, p<0,05), entretanto, a proliferação foi mais intensa no grupo 2 (1.262±440 cél /mm2) que no grupo 3 (653±300 cél /mm2, p<0,05 vs. group 2). O grau de apoptose encontrado foi semelhante entre o grupo 2 e o grupo 3. Houve aumento na expressão do p21 apenas no grupo 3 sendo que esta expressão foi semelhante nos grupos 1 e 2. Foi estudada também a resistência celular em túbulos proximais (TP) isolados de ratos normais (grupo Controle) e ratos submetidos à isquemia de 35 min, 24 h antes do estudo (grupo Isquemia). TP do grupo Isquemia foram susceptíveis à hipóxia, porém, resistentes à lesão de reoxigenação. Além disto, apresentaram menor produção de hidroperóxidos. Portanto, a resistência renal isquêmica aparentemente está associada a mecanismos celulares, o estresse oxidativo e o aumento na expressão do p21 são possíveis mediadores destes mecanismos. / Renal tubular resistance has been studied for the understanding of ischemic acute renal failure (ARF). Subsequent ischemic episodes may induce renal resistance whose mechanisms seem to be related to cell alterations. P21 is a cell cycle inhibitor that may be induced by oxygen free radicals and may have a protective effect in ischemic ARF. This study aimed at evaluating the role of oxidative stress and p21 on tubular resistance in isolated renal tubules and in a model of acquired resistance after renal ischemia. Wistar rats were divided into 3 groups: group 1 - sham; group 2 - submitted to sham procedure and after 2 days submitted to 45 min ischemia and group 3 - submitted to ischemia of 45 min followed by a second 45 min ischemia after 2 days. Plasma urea levels (114±60 vs. 136±44 mg/dL), serum creatinine (0.86±0.2 vs. 0.98±0.1mg/dL) and creatinine clearance (0.21±0.1vs. 0.24±0.1mL/min/100g.) evaluated at 48 hours after the second procedure were similar between groups 2 and 3 (all NS). ARF recovery time was also similar between groups 2 and 3. Histology disclosed the same degree of acute tubular necrosis between groups 2 and 3. Lymphocytes infiltrate was similar among all groups whereas macrophages infiltrate was greater in group 3. Enhanced cell proliferation was observed in groups 2 and 3 when compared with group 1 (125±28 cel/mm2, p<0.05), however it was greater in group 2 (1,262±440 cel/mm2) than group 3 (653±300 cel/mm2, p<0.05 vs. group 2). Degree of apoptosis was similar between groups 2 and 3. The p21 expression was increased only in group 3 whereas it was similar in groups 1 and 2. Cell resistance was also evaluated in isolated renal proximal tubules (PT) from control and ischemia groups. In the latter group, animals were submitted to 35 min ischemia and PT were isolated one day later. PT from the ischemia group were sensitive to hypoxia but resistant to reoxygenation injury which was followed by lower hydroperoxides production. In conclusion, renal resistance obtained by an ischemia was associated with cell mechanisms involving oxidative stress and increased p21 expression as mediators of this protection.

Isolated hip fracture following a minor fall is a serious injury, normally requiring urgent surgical treatment and a complex recovery journey. Although commonly associated with the elderly, incidence and impact in adults under 60 years of age may be underestimated. The extensive literature almost exclusively focuses on the elderly, surgical interventions and relatively short-term outcomes. Young adults are also missing from the dominant societal discourse and healthcare policy on fragility hip fracture. They therefore represent a silent sub-subset of the fragility hip fracture population, whose recovery experiences and needs, particularly in the longer term, remain largely unknown. A critical interpretivist approach and The Silences Framework (Serrant-Green, 2011), were used to ‘give voice’ to young adults with isolated hip fracture. Thirty participants, between one and ten years post injury, completed an in-depth, minimally structured interview in which they told their story of recovery. An inductive, thematic analysis was undertaken integrating Braun and Clarke (2006) and the four phase cyclical analysis of The Silences Framework (Serrant-Green, 2011). One cross-cutting theme: Communication emerged, together with four other main themes: Experience of care, Impact on self, Impact on others and Moving forward. 11 The findings indicated wide variation in the quality of care, often influenced by social and professional norms regarding hip fracture patient characteristics such as age and mode of injury. Multi-faceted, often long term, physical, social and psychological impact on participants, their family and wider social networks was also found. This included Post Traumatic Stress Disorder type symptoms and impact on work, finances and relationships. The study highlighted some limitations of the current hip fracture care pathway for supporting the specific recovery needs of young adults. It also identified some limited effectiveness of commonly used patient reported outcome measures for hip fracture in this young client group. Exploring the recovery experiences of this under-represented group confirmed, but also altered the silences initially identified. Furthermore, it uncovered new silences which informed recommendations for future research; healthcare practice and policy. This study offers the first long term exploration of the impact of isolated hip fracture following a minor fall in young adults from their perspective. In doing so, it has also demonstrated the appropriateness of The Silences Framework (Serrant-Green, 2011) for guiding a person-centred, experience-based, acute orthopaedic/rehabilitation study undertaken by a student researcher.

Dans cette thèse, on présente de nouvelles approches à base de parcimonie et d'invariance d' échelle pour le développement de techniques rapides et efficaces en traitement d'images. Au lieu d'utiliser la norme l1 pour imposer la parcimonie, on exploite plutôt des pénalités non-convexes qui encouragent plus la parcimonie. On propose une approche de premier ordre pour estimer une solution d'un opérateur proximal non-convexe, ce qui permet d'exploiter facilement la non-convexité. On étudie aussi le problème de pluri-parcimonie quand le problème d'optimisation est composé de plusieurs termes parcimonieux. Ce cas survient généralement dans les problèmes qui nécessitent à la fois une estimation robuste pour rejeter les valeurs aberrantes et exploiter une information de parcimonie connue a priori. Ces techniques sont appliquées à plusieurs problèmes importants en vision par ordinateur bas niveau telles que le lissage sélectif, la séparation d'images, l'intégration robuste et la déconvolution. On propose aussi d'aller au-delà de la parcimonie et apprendre un modèle de mapping spectral non-local pour le débruitage d'images. La notion d'invariance d' échelle joue aussi un rôle important dans nos travaux. En exploitant ce principe, une définition précise des contours est définie, ce qui peut être complémentaire à la notion de parcimonie. Plus précisément, on peut construire des représentations invariantes pour la classification en se basant sur une architecture de réseaux convolutionnels profonds. L'invariance d' échelle permet aussi d'extraire les pixels qui portent les informations nécessaires pour la reconstruction ou aussi améliorer l'estimation du flot optique sur les images turbulentes en imposant la parcimonie comme régularisation sur les exposants de singularité locaux. / In this thesis, we present new techniques based on the notions of sparsity and scale invariance to design fast and efficient image processing applications. Instead of using the popular l1-norm to model sparsity, we focus on the use of non-convex penalties that promote more sparsity. We propose to use a first-order approximation to estimate a solution of non-convex proximal operators, which permits to easily use a wide rangeof penalties. We address also the problem of multi-sparsity, when the minimization problem is composed of various sparse terms, which typically arises in problems that require both a robust estimation to reject outliers and a sparse prior. These techniques are applied to various important problems in low-level computer vision such as edgeaware smoothing, image separation, robust integration and image deconvolution. We propose also to go beyond sparsity models and learn non-local spectral mapping with application to image denoising. Scale-invariance is another notion that plays an important role in our work. Using this principle, a precise definition of edges can be derived which can be complementary to sparsity. More precisely, we can extractinvariant features for classification from sparse representations in a deep convolutional framework. Scale-invariance permits also to extract relevant pixels for sparsifying images. We use this principle as well to improve optical ow estimation on turbulent images by imposing a sparse regularization on the local singular exponents instead of regular gradients.

Une approche efficace pour la résolution de problèmes inverses consiste à définir le signal (ou l'image) recherché(e) par minimisation d'un critère pénalisé. Ce dernier s'écrit souvent sous la forme d'une somme de fonctions composées avec des opérateurs linéaires. En pratique, ces fonctions peuvent n'être ni convexes ni différentiables. De plus, les problèmes auxquels on doit faire face sont souvent de grande dimension. L'objectif de cette thèse est de concevoir de nouvelles méthodes pour résoudre de tels problèmes de minimisation, tout en accordant une attention particulière aux coûts de calculs ainsi qu'aux résultats théoriques de convergence. Une première idée pour construire des algorithmes rapides d'optimisation est d'employer une stratégie de préconditionnement, la métrique sous-jacente étant adaptée à chaque itération. Nous appliquons cette technique à l'algorithme explicite-implicite et proposons une méthode, fondée sur le principe de majoration-minimisation, afin de choisir automatiquement les matrices de préconditionnement. L'analyse de la convergence de cet algorithme repose sur l'inégalité de Kurdyka-L ojasiewicz. Une seconde stratégie consiste à découper les données traitées en différents blocs de dimension réduite. Cette approche nous permet de contrôler à la fois le nombre d'opérations s'effectuant à chaque itération de l'algorithme, ainsi que les besoins en mémoire, lors de son implémentation. Nous proposons ainsi des méthodes alternées par bloc dans les contextes de l'optimisation non convexe et convexe. Dans le cadre non convexe, une version alternée par bloc de l'algorithme explicite-implicite préconditionné est proposée. Les blocs sont alors mis à jour suivant une règle déterministe acyclique. Lorsque des hypothèses supplémentaires de convexité peuvent être faites, nous obtenons divers algorithmes proximaux primaux-duaux alternés, permettant l'usage d'une règle aléatoire arbitraire de balayage des blocs. L'analyse théorique de ces algorithmes stochastiques d'optimisation convexe se base sur la théorie des opérateurs monotones. Un élément clé permettant de résoudre des problèmes d'optimisation de grande dimension réside dans la possibilité de mettre en oeuvre en parallèle certaines étapes de calculs. Cette parallélisation est possible pour les algorithmes proximaux primaux-duaux alternés par bloc que nous proposons: les variables primales, ainsi que celles duales, peuvent être mises à jour en parallèle, de manière tout à fait flexible. A partir de ces résultats, nous déduisons de nouvelles méthodes distribuées, où les calculs sont répartis sur différents agents communiquant entre eux suivant une topologie d'hypergraphe. Finalement, nos contributions méthodologiques sont validées sur différentes applications en traitement du signal et des images. Nous nous intéressons dans un premier temps à divers problèmes d'optimisation faisant intervenir des critères non convexes, en particulier en restauration d'images lorsque l'image originale est dégradée par un bruit gaussien dépendant du signal, en démélange spectral, en reconstruction de phase en tomographie, et en déconvolution aveugle pour la reconstruction de signaux sismiques parcimonieux. Puis, dans un second temps, nous abordons des problèmes convexes intervenant dans la reconstruction de maillages 3D et dans l'optimisation de requêtes pour la gestion de bases de données / An efficient approach for solving an inverse problem is to define the recovered signal/image as a minimizer of a penalized criterion which is often split in a sum of simpler functions composed with linear operators. In the situations of practical interest, these functions may be neither convex nor smooth. In addition, large scale optimization problems often have to be faced. This thesis is devoted to the design of new methods to solve such difficult minimization problems, while paying attention to computational issues and theoretical convergence properties. A first idea to build fast minimization algorithms is to make use of a preconditioning strategy by adapting, at each iteration, the underlying metric. We incorporate this technique in the forward-backward algorithm and provide an automatic method for choosing the preconditioning matrices, based on a majorization-minimization principle. The convergence proofs rely on the Kurdyka-L ojasiewicz inequality. A second strategy consists of splitting the involved data in different blocks of reduced dimension. This approach allows us to control the number of operations performed at each iteration of the algorithms, as well as the required memory. For this purpose, block alternating methods are developed in the context of both non-convex and convex optimization problems. In the non-convex case, a block alternating version of the preconditioned forward-backward algorithm is proposed, where the blocks are updated according to an acyclic deterministic rule. When additional convexity assumptions can be made, various alternating proximal primal-dual algorithms are obtained by using an arbitrary random sweeping rule. The theoretical analysis of these stochastic convex optimization algorithms is grounded on the theory of monotone operators. A key ingredient in the solution of high dimensional optimization problems lies in the possibility of performing some of the computation steps in a parallel manner. This parallelization is made possible in the proposed block alternating primal-dual methods where the primal variables, as well as the dual ones, can be updated in a quite flexible way. As an offspring of these results, new distributed algorithms are derived, where the computations are spread over a set of agents connected through a general hyper graph topology. Finally, our methodological contributions are validated on a number of applications in signal and image processing. First, we focus on optimization problems involving non-convex criteria, in particular image restoration when the original image is corrupted with a signal dependent Gaussian noise, spectral unmixing, phase reconstruction in tomography, and blind deconvolution in seismic sparse signal reconstruction. Then, we address convex minimization problems arising in the context of 3D mesh denoising and in query optimization for database management

Introduction: The forearm is a complex biological unit, which has allowed man's evolution. This PhD commenced with an analysis of the normal biomechanical functioning of the key components of the forearm: notably the distal radioulnar joint (DRUJ), interosseous ligament (IOL) and proximal radioulnar joint (PRUJ). Understanding normal forearm physiology, a clinical study followed to delineate the pathophysiology of a new clinical entity, related to DRUJ dysfunction. Methods: Biomechanical Study: A biomechanical testing jig was developed to facilitate collection of data about normal functioning of the DRUJ, IOL and PRUJ in both unloaded and loaded states. This permitted testing throughout the range of forearm pronosupination. Thawed fresh frozen cadaveric upperlimbs were mounted into the jig. Using Microstrain® strain gauges and Tekscan™ pressure sensors, the functional anatomy of the key components of the forearm was delineated, both with the forearm flexed at 90° and maximally extended at the elbow. Clinical Study: A series of 3-Tesla MRI scans was undertaken on patients symptomatic of an intermittent ulnar neuropathy. The causative pathophysiology was determined using 3D qualitative and quantitative analyses. Results: Biomechanical Study: Reproducible patterns of force transmitted and joint contact area have been determined for the DRUJ, and for the first time, the PRUJ. With the exception of PMax and P60 for the PRUJ, application of load increases contact areas and transmitted forces across the joints (P<0.05). The converse is true for PMax and P60 in the PRUJ. The IOL is lax during pronation, strain gradually increasing as the arm moves to neutral. In neutral the middle-portion of the IOL (m-IOL) demonstrates most strain, this decreasing again in supination, whilst the distal and proximal portions (d- & p-IOL) exhibit more strain (P<0.05). Axial loading consistently increases strain in all ligaments (P<0.05). Observed behaviour patterns across the joints and in the ligaments alter with elbow extension (P<0.05). Clinical Study: Salient symptoms of the new syndrome were described. Displacement of the ulnar nerve from its normal course was seen with compression/distraction in the distal forearm and Guyon’s canal. This was considered causative of the syndrome. As a by-product of the research, a new clinical device was also developed, which improves the patient pathway when investigating DRUJ dysfunction. Conclusions and Outcomes: This research has analysed normal forearm biomechanics determining that the PRUJ is a load-bearing joint, interrelated with the DRUJ and IOL. Elbow extension has been shown to alter the normal biomechanics of the forearm. A clinical entity of a dysfunctional forearm has been defined, called subluxation-related ulnar neuropathy or SUN syndrome. Finally, a new clinical device has been developed, which it is anticipated will translate into visible improvements in patient care.

O arroz (Oryza sativa L.) é um dos cereais mais produzidos e consumidos no mundo, caracterizando como principal alimento para mais da metade da população mundial. Sua importância é destacada principalmente em países em desenvolvimento, tais como o Brasil, desempenhando papel estratégico em níveis econômico e social. Atualmente, é crescente o número de pesquisas que integram desde o melhoramento vegetal e o manejo da adubação até o aproveitamento dos nutrientes pelo organismo humano. O melhoramento de plantas pode ajudar a melhorar a dieta humana, com o desenvolvimento de plantas com maiores teores de vitaminas e micronutrientes. A introdução de produtos agrícolas biofortificados como o arroz, complementará as intervenções em nutrição existentes. Sendo assim, tornase imprescindível conhecer como as condições de cultivo influenciam na qualidade do grão e na importância para obtenção de um alimento com maior valor biológico e nutricional e com menores concentrações de antinutricionais. O objetivo deste trabalho foi analisar as características físico-químicas e nutricionais do arroz (Oryza sativa L.) integral, promovidas por diferentes adubações (experimento I) e aos grãos biofortificados (experimento II). Foram realizadas análises de composição centesimal, composição mineral, diálise in vitro de minerais, vitaminas do complexo B, conteúdo de compostos fenólicos e fitatos em grão, além das análises de parâmetros de qualidade em arroz como conteúdo de amilose aparente, propriedade de pasta e análise instrumental de cor. Os resultados obtidos no estudo assemelham-se aos citados nas literaturas quanto á composição centesimal e mineral, enfatizando a influência da cocção na composição nutricional do grão e em seus compostos como fenólicos e fitatos. Para o dialisado de cálcio e zinco no arroz biofortificado, a variação nos valores foi em média de 1,14% a 3,26% para cálcio e 0,80% a 3,43% para zinco. No experimento II, as perdas de vitaminas para o arroz integral biofortificado após o cozimento foi em torno de 33,4%, 52,17% e 76% para vitamina B1, vitamina B2 e vitamina B6, respectivamente. Por fim, conclui-se que a produção de variedades melhoradas seja através de adubações ou da escolha de melhores cultivares, que apresentam maior teor de minerais e vitaminas, foram eficientes para a melhora da disponibilidade nutricional dos grãos. / Rice (Oryza sativa L.) is one of most cereals produced and consumed in the world, characterized as the primary food for more than half the world population. Its importance is highlighted especially in developing countries such as Brazil, playing a strategic role in economic and social levels. There is increasing the number of studies since that integrate plant breeding and fertilizer management to use the nutrients the human body. Plant breeding can help improve the human diet, with the development of plants with higher levels of vitamins and micronutrients. The introduction of biofortified crops like rice, will complement existing nutrition interventions. Thus, it becomes essential to know as the cultivation conditions influence the quality of grain and the importance to obtain a food with higher nutritional and biological value and with lower concentrations of antinutritional. Threfore, the aim of this study was to analyze the pshycochemical and nutritional properties of rice (Oryza sativa L.) in full, caused by different fertilization (experiment I) and subjected to biofortification (experiment II). Were performed to analyze chemical composition, mineral, dialysis in vitro of minerals, vitamins B, content of phenolic compounds and phytates in grain, in addition to the analysis of quality parameters in rice as apparent amylase content, folder property and instrumental analysis of color. The results obtained in this study are similar to those cited in other literature as to the mineral and chemical composition, emphasizing the influence of cooking on the nutritional composition of grain and as phenolic compounds and phytates. For the dialysate calcium and zinc in rice biofortified crops, the variation in the values averaged 1.14% to 3.26% for calcium and 0.80% to 3.43% for zinc. Losses of vitamins to the rice after cooking was around 33.4%, 52.17% and 76% for vitamin B1, vitamin B2 and vitamin B6, respectively. Finally, we conclude that the production of improved varieties is through fertilization or the best choice of cultivars, that have a higher content of minerals and vitamins, were effective to improve the availability of nutrients of the grains.

La pathogénie du syndrome néphrotique idiopathique est inconnue, mais de nombreux arguments clinques et expérimentaux favorisent l’hypothèse d’une pathogénie dys-immunitaire à expression immunologique et rénale, au cours de laquelle on observerait une altération des lymphocytes T. Cependant, le mécanisme exact reste encore mal connu. Récemment, le Rituximab, un anticorps dirigé contre l’antigène CD20, a montré une efficacité à induire une rémission à moyen et long terme suggérant l’implication d’une dysfonction des lymphocytes B et/ou un défaut de coopération T-B. Notre laboratoire a isolé un nouveau gène C-MIP dont l’expression est induite dans certaines sous-populations lymphocytaires T et B, ainsi que dans les podocytes de patients atteints de SNI en phase de poussée mais quasiment indétectable chez les sujets sains.Dans ces travaux, ancillaires au PHRC NEPHRUTIX, nous avons étudié les perturbations lymphocytaires T, avant, au moment de la rechute et en période de rémission au cours de syndrome néphrotique à lésions glomérulaires minimes et l’effet du traitement par le Rituximab. Dans cette étude, nous avons mis en évidence que la rechute était associée à un effondrement des lymphocytes T régulateurs, une baisse profonde de l’interleukine-2 ainsi qu’à une surexpression significative de C-MIP, précédant la survenue de la rechute. Ces modifications se restaurent en rémission. Enfin, la rémission obtenue dans le bras Rituximab, entraîne une diminution des lymphocytes T folliculaires (Tfh), des iNKT et des cellules double-négatives DN-TCR Vα24, suggérant que le SNLGM implique un défaut des réponses immunitaires innées et adaptatives, qui peut être stabilisé par un traitement par Rituximab.Afin d’étudier le rôle de C-MIP, nous avons généré des souris transgéniques sur-exprimant ce gène dans les lymphocytes T matures périphériques. Cette surexpression est à l’origine d’un phénotype lymphocytaire altéré marqué par une accumulation de lymphocytes T naïfs, un effondrement des cytokines activatrices de type Th1 et Th2 et une accumulation des formes inactives des Src kinases. Ces résultats suggèrent que C-MIP, en inhibant les Src kinases, est un régulateur négatif de l’activation T impliqué dans la signalisation proximale et pourrait être impliqué dans l’hypo-réactivité lymphocytaire T observée chez les patients atteints de SNLGM actif. / The pathogenesis of minimal-change nephrotic syndrom (MCNS) is unknown, but, supported by many clinical and experimental arguments, it was suggested that MCNS is a dys-immune disorder with immunogical and renal expression, during which T-cell alteration would be observed. However, the exact mechanism remains unknown. Recently, Rituximab, a B-cell depleting agent, is effctive in inducing mid- and long-term remission suggesting involvement of B-cell dysfunction and/or lack of T-B cooperation. Our laboratory identified a new gene: C-MIP. We have shown that C-MIP abundance is increased in some T and B lymphocyte subpopulations, as well as in podocytes of MCNS patients during relapse phase but undetectable in healthy subjects.In this work, ancillary to the NEPHRUTIX PHRC, we studied T-cell disturbances before and during the relapse or during the remission time in MCNS and the effect of Rituximab therapy. In this study, we found that relapses were associated with significant decrease in regulatory T cell and interleukin-2 expression, while C-MIP abundance was significantly increased. These changes are restored during remission time. Finally, remission after Rituximab therapy leads to a decrease in follicular T cells (Tfh), iNKT and double-negative (CD4- CD8-) T cells expressing the invariant Vα24 chain, suggesting that MCNS involves a disorder of innate and adaptative immune response, which can be stabilized by Rituximab treatment.In order to study the C-MIP role, we generated transgenic mice overexpressing this gene in the peripheral mature T-cells. This overexpression leads to an altered lymphocyte phenotype with an accumulation of naive T lymphocytes, a significant decrease of Th1 and Th2 activating cytokines and accumulation of inactive Src kinases. These results suggest that, by inhibiting Src kinases, C-MIP is a negative regulator of activation T involved in proximal signalling and may be responsable of the lymphocyte T hypo-reactivity observed in patients with active MCNS.

Orientador: Enner Herenio de Alcântara / Resumo: O Material Particulado em Suspensão (MPS) é o principal componente em sistemas aquáticos. Elevadas concentrações de MPS implicam na atenuação da luz, e ocasionam alterações das taxas fotossintéticas. Além disso, a presença de MPS no sistema aquático pode aumentar os níveis de turbidez, absorver poluentes e podem ser considerados como um indicativo de descargas de escoamento superficial. Portanto, monitorar as concentrações de MPS é essencial para a gerar informações técnicas que subsidiem o correto manejo dos recursos aquáticos, prevenindo colapsos hidrológicos. O sensoriamento remoto se mostra como uma eficiente ferramenta para monitorar e mapear MPS quando comparada às técnicas tradicionais de monitoramento, como as medidas in situ. Entretanto, diante de uma grande e complexa variabilidade de componentes óticos, desenvolver modelos de MPS por meio do sinal registrado em sensores remotos é um desafio. Diversos modelos foram desenvolvidos para reservatórios, lagos e lagoas específicos. Atualmente, não há um único modelo capaz de estimar MPS em reservatórios brasileiros em cascata. Com o objetivo de estimar as concentrações de MPS de forma acurada, o objetivo desta tese foi desenvolver um modelo semi-analítico capaz de estimar valores de coeficiente de atenuação, Kd, por meio do uso dos coeficientes de absorção e espalhamento e, consequentemente, utilizar os valores de Kd para estimar as concentrações de MPS. A adoção desta estratégica se baseou na atenuação da luz ao longo da... (Resumo completo, clicar acesso eletrônico abaixo) / Abstract: Suspended particulate matter (SPM) is the main component presented within aquatic system. High levels of SPM concentration attenuate the light affecting the photosynthesis rates. Besides, can increase turbidity levels, absorb pollutions and is an indicative of runoff discharges. Therefore, monitoring SPM concentrations is essential to provide reliable information for a correct water management to prevent hydrological collapse. Remote sensing emerges as an efficient tool to map and monitor SPM when compared to traditional techniques, such as in situ measurements. Nevertheless, considering a widely range of optical components, modeling the remote sensing signal in terms of SPM is a challenge. Several models were developed for specific reservoirs, lakes or ponds. Up to our knowledge, there is not a single model capable to retrieve SPM in Brazilian linked reservoirs in a cascade system. In order to accurately estimate SPM, the aim of the thesis was developed a semianalytical model capable to estimate Kd via absorption and backscattering coefficients, and then, use Kd to derive SPM. This approach was adopted because SPM directly contributes to the light attenuation within the water column. Firstly, optical features were investigated. It was found that each reservoir presented a specific optical active component (OAC) dominance, such as Barra Bonita, the first reservoir in cascade is dominated by organic SPM, while Nova Avanhandava, the last reservoir in cascade is dominated by ino... (Complete abstract click electronic access below) / Doutor

Les problèmes inverses en traitement d'images peuvent être résolus en utilisant des méthodes variationnelles classiques, des approches basées sur l'apprentissage profond, ou encore des stratégies bayésiennes. Bien que différentes, ces approches nécessitent toutes des algorithmes d'optimisation efficaces. L'opérateur proximal est un outil important pour la minimisation de fonctions non lisses. Dans cette thèse, nous illustrons la polyvalence des algorithmes proximaux en les introduisant dans chacune des trois méthodes de résolution susmentionnées.Tout d'abord, nous considérons une formulation variationnelle sous contraintes dont la fonction objectif est composite. Nous développons PIPA, un nouvel algorithme proximal de points intérieurs permettant de résoudre ce problème. Dans le but d'accélérer PIPA, nous y incluons une métrique variable. La convergence de PIPA est prouvée sous certaines conditions et nous montrons que cette méthode est plus rapide que des algorithmes de l'état de l'art au travers de deux exemples numériques en traitement d'images.Dans une deuxième partie, nous étudions iRestNet, une architecture neuronale obtenue en déroulant un algorithme proximal de points intérieurs. iRestNet nécessite l'expression de l'opérateur proximal de la barrière logarithmique et des dérivées premières de cet opérateur. Nous fournissons ces expressions pour trois types de contraintes. Nous montrons ensuite que sous certaines conditions, cette architecture est robuste à une perturbation sur son entrée. Enfin, iRestNet démontre de bonnes performances pratiques en restauration d'images par rapport à une approche variationnelle et à d'autres méthodes d'apprentissage profond.La dernière partie de cette thèse est consacrée à l'étude d'une méthode d'échantillonnage stochastique pour résoudre des problèmes inverses dans un cadre bayésien. Nous proposons une version accélérée de l'algorithme proximal de Langevin non ajusté, baptisée PP-ULA. Cet algorithme est incorporé à un échantillonneur de Gibbs hybride utilisé pour réaliser la déconvolution et la segmentation d'images ultrasonores. PP-ULA utilise le principe de majoration-minimisation afin de gérer les distributions non log-concaves. Comme le montrent nos expériences réalisées sur des données ultrasonores simulées et réelles, PP-ULA permet une importante réduction du temps d'exécution tout en produisant des résultats de déconvolution et de segmentation très satisfaisants. / Inverse problems in image processing can be solved by diverse techniques, such as classical variational methods, recent deep learning approaches, or Bayesian strategies. Although relying on different principles, these methods all require efficient optimization algorithms. The proximity operator appears as a crucial tool in many iterative solvers for nonsmooth optimization problems. In this thesis, we illustrate the versatility of proximal algorithms by incorporating them within each one of the aforementioned resolution methods.First, we consider a variational formulation including a set of constraints and a composite objective function. We present PIPA, a novel proximal interior point algorithm for solving the considered optimization problem. This algorithm includes variable metrics for acceleration purposes. We derive convergence guarantees for PIPA and show in numerical experiments that it compares favorably with state-of-the-art algorithms in two challenging image processing applications.In a second part, we investigate a neural network architecture called iRestNet, obtained by unfolding a proximal interior point algorithm over a fixed number of iterations. iRestNet requires the expression of the logarithmic barrier proximity operator and of its first derivatives, which we provide for three useful types of constraints. Then, we derive conditions under which this optimization-inspired architecture is robust to an input perturbation. We conduct several image deblurring experiments, in which iRestNet performs well with respect to a variational approach and to state-of-the-art deep learning methods.The last part of this thesis focuses on a stochastic sampling method for solving inverse problems in a Bayesian setting. We present an accelerated proximal unadjusted Langevin algorithm called PP-ULA. This scheme is incorporated into a hybrid Gibbs sampler used to perform joint deconvolution and segmentation of ultrasound images. PP-ULA employs the majorize-minimize principle to address non log-concave priors. As shown in numerical experiments, PP-ULA leads to a significant time reduction and to very satisfactory deconvolution and segmentation results on both simulated and real ultrasound data.

Perfusion magnetic resonance imaging is a medical diagnostic method which requires high spatial and temporal resolution simultaneously to capture dynamics of an intravenous contrast agent which is used to perfusion measurement. However, magnetic resonance imaging has physical limits which do not allow to have this resolution simultaneously. This thesis deals with compressed sensing which enables to reconstruct measured data from relatively few acquired samples (below Nyquist rate) while resolution required to perfusion analysis is increased. This aim could be achieved with suitably proposed apriory information about sensed data and model proposal. The reconstruction is then done as an optimization problem. Doctoral thesis brings several new reconstruction models, further proposes method to debias this estimates and examines influence of compressed sensing onto perfusion parameters. Whole thesis is ended with extension of compressed sensing into three-dimensional data. Here, the influence of reconstruction onto perfusion parameters is also described. In summary, the thesis shows that due to compressed sensing, temporal resolution can be increased with the fixed spatial resolution or spatial resolution can be increased with the fixed temporal resolution.