Les charges que subit la voilure d’un avion ne peuvent pas être directement mesurées en vol. Ces charges sont leplus souvent estimées à partir des déformations de la voilure, mesurées par des ponts de jauge. La relation entreles déformations et les charges est typiquement modélisée par un modèle de régression linéaire. L'estimation descharges en vol est ainsi réalisée par une méthode en 2 étapes, connue sous le nom de méthode de Skopinski :- l'étalonnage au sol : des essais sont réalisés pour identifier les paramètres du modèle reliant les déformations,mesurées au sol, aux charges, connues à partir des efforts appliqués sur la structure.- les essais en vol : les charges sont estimées à partir des déformations, mesurées en vol, et des paramètres dumodèle identifiés au sol.Dans cette méthode, les incertitudes existant à chaque étape ne sont pas prises en compte. Ces incertitudescorrespondent aux bruits de mesure mais aussi aux erreurs de modélisation. De plus, le domaine d'application dumodèle est différent du domaine dans lequel ses paramètres sont identifiés. En effet, le modèle est étalonné au soldans des conditions de pression, de température et de chargement différentes des conditions existant en vol. Le butde cette thèse est de développer une méthode permettant de prendre en compte ces différentes sourcesd’incertitude afin, d’une part, de mieux identifier le modèle et, d’autre part, de quantifier l’incertitude qu’il entraînelors de son utilisation. / The loads on the wings of aircraft cannot be directly measured in flight. These loads are most of the time estimatedfrom the strains of the wing, which are measured by strain gages bridges. The relation between the strains and theloads is typically modeled by a linear regression model. The estimation of flight loads is so performed by a methodin 2 steps, known as the Skopinski method :- the ground calibration : tests are performed in order to identify the model parameters linking the strains, measuredon ground, to the loads, known from the loads which are applied on the structure.- the flight tests : the loads are estimated from the strains, measured in flight, and from the model parameters,identified on ground.In this method, the existing uncertainties at each step are not taken into account. These uncertainties correspond tothe measurement noises and the modeling errors. Furthermore, the model is applied in a domain which is differentfrom the domain where its parameters are identified. Indeed, the model is calibrated on ground in pressure, thermaland loading conditions which are different from those existing in flight. The aim of this PhD is to develop a methodtaken into account these different sources of uncertainties to better identify the model on one hand and to quantifythe uncertainty which is caused by its use.

Em radioterapia, o espectro de energia é o componente mais importante dos feixes de elétrons. Espectros de energia de elétrons são relevântes para o cálculo acurado da dose, aplicações do sistema de planejamento e simulações realistas. Reconstrução inversa consiste na derivação do espectro de energia de elétrons a partir de curvas de porcentagem de dose em profundidade utilizando um apropiado modelo matemático. Reconstrução inversa é considerada a melhor dentre muitas abordagens porque: i) não requer nenhum equipamento suplementar ou do conhecimento detalhado da geometria e composição do cabeçote do acelerador; ii) equipamentos para a medição de curvas de porcentagem de dose em profundidade estão disponíveis em qualquer clínica e iii) é computacionalmente rápida. Neste trabalho, usou-se o método de reconstrução inversa baseado na sinergia recozimento simulado generalizado-regularização de Tikhonov. A validação da reconstrução foi realizada através do índice gama sob critérios clínicos de aceitação restritivos. Resultados mostraram que os espectros de energia reconstruídos reproduzem com precisão a porcentagem de dose em profundidade clínica bem como valores de dose fora do eixo central. Assim, concluí-se que o método empregado é ecaz para reconstruir espectros de energia que representam efetivamente espectros de energia do acelerador que atingem na supercie do fantoma. Consequentemente, sob certos limites, eles poderiam auxiliar em simulações realistas do tratamento. / In radiotherapy, energy spectrum is the most critical component of any electron beam. Knowledge of energy spectrum is important for accurate dose calculation, treatment planning applications and realistic simulations. Inverse reconstruction derives energy spectrum from the measured percentage depth dose using an appropriate mathematical model. There are several advantages to using inverse reconstruction: i) it does not require any supplementary equipment or detailed knowledge of the geometry head and composition; ii) the equipment for measurement of the percentage depth dose is standard and already available in any clinic and iii) it is computationally fast. In this work, we used the inverse reconstruction method based on the synergy simulated annealing generalized-Tikhonov regularization. Validation of inverse reconstruction was done by comparing the measured and reconstructed percentage depth dose via the gamma index. Results show the reconstructed electron energy spectra accurately reproduce the clinical dose percentage as well as o-axis dose values. Therefore, it was concluded that the method employed is eective to reconstruct energy spectra that eectively represent accelerator energy spectra reaching the phantom surface. Consequently, under certain limits, they could aid in realistic simulations of treatment.

L’imagerie de source corticale joue un rôle important pour la compréhension fonctionnelle ou pathologique du cerveau. Elle permet d'estimer l'activation de certaines zones corticales en réponse à un stimulus cognitif donné et elle est également utile pour identifier la localisation des activités pathologiques, qui sont les premières étapes de l'étude des activations de réseaux neuronaux sous-jacents. Diverses méthodes d'investigation clinique peuvent être utilisées, des modalités d'imagerie (TEP, IRM) et magnéto-électroencéphalographie (EEG, SEEG, MEG). Nous souhaitions résoudre le problème à partir de données non invasives : les mesures de l'EEG de scalp, elle procure une résolution temporelle à la hauteur des processus étudiés Cependant, la localisation des sources activées à partir d'enregistrements EEG reste une tâche extrêmement difficile en raison de la faible résolution spatiale. Pour ces raisons, nous avons restreint les objectifs de cette thèse à la reconstruction de cartes d’activation des sources corticales de surface. Différentes approches ont été explorées. Les méthodes les plus simples d'imagerie corticales sont basées uniquement sur les caractéristiques géométriques de la tête. La charge de calcul est considérablement réduite et les modèles utilisés sont faciles à mettre en œuvre. Toutefois, ces approches ne fournissent pas d'informations précises sur les générateurs neuronaux et sur leurs propriétés spatiotemporelles. Pour surmonter ces limitations, des techniques plus sophistiquées peuvent être utilisées pour construire un modèle de propagation réaliste, et donc d'atteindre une meilleure reconstruction de sources. Cependant, le problème inverse est sévèrement mal posé, et les contraintes doivent être imposées pour réduire l'espace des solutions. En l'absence de modèle bioanatomique, les méthodes développées sont fondées sur des considérations géométriques de la tête ainsi que la propagation physiologique des sources. Les opérateurs matriciels de rang plein sont appliqués sur les données, de manière similaire à celle effectuée par les méthodes de surface laplacien, et sont basés sur l'hypothèse que les données de surface peuvent être expliquées par un mélange de fonctions de bases radiales linéaires produites par les sources sous-jacentes. Dans la deuxième partie de ces travaux, nous détendons la contrainte-de rang plein en adoptant un modèle de dipôles distribués sur la surface corticale. L'inversion est alors contrainte par une hypothèse de parcimonie, basée sur l'hypothèse physiologique que seuls quelques sources corticales sont simultanément actives ce qui est particulièrement valable dans le contexte des sources d'épilepsie ou dans le cas de tâches cognitives. Pour appliquer cette régularisation, nous considérons simultanément les deux domaines spatiaux et temporels. Nous proposons deux dictionnaires combinés d’atomes spatio-temporels, le premier basé sur une analyse en composantes principales des données, la seconde à l'aide d'une décomposition en ondelettes, plus robuste vis-à-vis du bruit et bien adaptée à la nature non-stationnaire de ces données électrophysiologiques. Toutes les méthodes proposées ont été testées sur des données simulées et comparées aux approches classiques de la littérature. Les performances obtenues sont satisfaisantes et montrent une bonne robustesse vis-à-vis du bruit. Nous avons également validé notre approche sur des données réelles telles que des pointes intercritiques de patients épileptiques expertisées par les neurologues de l'hôpital universitaire de Nancy affiliées au projet. Les localisations estimées sont validées par l'identification de la zone épileptogène obtenue par l'exploration intracérébrale à partir de mesures stéréo EEG. / Cortical Source Imaging plays an important role for understanding the functional and pathological brain mechanisms. It links the activation of certain cortical areas in response to a given cognitive stimulus, and allows one to study the co-activation of the underlying functional networks. Among the available acquisition modality, electroencephalographic measurements (EEG) have the great advantage of providing a time resolution of the order of the millisecond, at the scale of the dynamic of the studied process, while being a non-invasive technique often used in clinical routine. However the identification of the activated sources from EEG recordings remains an extremely difficult task because of the low spatial resolution this modality provides, of the strong filtering effect of the cranial bones and errors inherent to the used propagation model. In this work different approaches for the estimation of cortical activity from surface EEG have been explored. The simplest cortical imaging methods are based only on the geometrical characteristics of the head. The computational load is greatly reduced and the used models are easy to implement. However, such approaches do not provide accurate information about the neural generators and on their spatiotemporal properties. To overcome such limitations, more sophisticated techniques can be used to build a realistic propagation model, and thus to reach better source reconstruction by its inversion. However, such inversion problem is severely ill-posed, and constraints have to be imposed to reduce the solution space. We began by reconsidering the cortical source imaging problem by relying mostly on the observations provided by the EEG measurements, when no anatomical modeling is available. The developed methods are based on simple but universal considerations about the head geometry as well as the physiological propagation of the sources. Full-rank matrix operators are applied on the data, similarly as done by Surface Laplacian methods, and are based on the assumption that the surface can be explained by a mixture of linear radial basis functions produced by the underlying sources. In the second part of the thesis, we relax the full-rank constraint by adopting a distributed dipole model constellating the cortical surface. The inversion is constrained by an hypothesis of sparsity, based on the physiological assumption that only a few cortical sources are active simultaneously Such hypothesis is particularly valid in the context of epileptic sources or in the case of cognitive tasks. To apply this regularization, we consider simultaneously both spatial and temporal domains. We propose two combined dictionaries of spatio-temporal atoms, the first based on a principal components analysis of the data, the second using a wavelet decomposition, more robust to noise and well suited to the non-stationary nature of these electrophysiological data. All of the proposed methods have been tested on simulated data and compared to conventional approaches of the literature. The obtained performances are satisfactory and show good robustness to the addition of noise. We have also validated our approach on real epileptic data provided by neurologists of the University Hospital of Nancy affiliated to the project. The estimated locations are consistent with the epileptogenic zone identification obtained by intracerebral exploration based on Stereo-EEG measurements.

Submitted by Maria Cristina (library@lncc.br) on 2015-10-08T15:17:16Z No. of bitstreams: 1 tesi_Jemy.pdf: 3188495 bytes, checksum: 6f19a9e47b46b25dd3bf6e8713033c41 (MD5) / Approved for entry into archive by Maria Cristina (library@lncc.br) on 2015-10-08T15:17:27Z (GMT) No. of bitstreams: 1 tesi_Jemy.pdf: 3188495 bytes, checksum: 6f19a9e47b46b25dd3bf6e8713033c41 (MD5) / Made available in DSpace on 2015-10-08T15:17:37Z (GMT). No. of bitstreams: 1 tesi_Jemy.pdf: 3188495 bytes, checksum: 6f19a9e47b46b25dd3bf6e8713033c41 (MD5) Previous issue date: 2015-03-03 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / The aim of this dissertation is to evaluate the value of certain parameters of a partial differential equation, using a Iterative method of Regularization (Landweber Nonlinear). This problem is motivated by the behavior of ion channels of the cell neural, which is of difficult experimental determination. We use a simplified model, in the case the passive cable equation, which is a linear parabolic differential equation, with terms of diffusion and reaction, not necessarily homogeneous. We consider that the terms of reaction are given by a function that depends on the variable space, and are unknown. To determine this function we use the nonlinear Landweber method, that, in a Hilbert space, search iterative approximations for the unknown function. Each step of this algorithm requires the solution of two parabolic partial differential equations and an integral and use the finite difference method to obtain the approximate solution of the partial differential equations, and the trapezoid method to obtain the solution of the integral, resulting a method computationally intensive. In this dissertation we describe the biological motivation of the problem and the mathematical basis of the algorithm, and teste various computer cases. / O objetivo desta dissertação é obter de forma indireta o valor de certos parâmetros de uma equação diferencial parcial, utilizando um método de Regularização Iterativo (Landweber não Linear). Este problema é motivado pelo comportamento de canais iônicos da célula neuronal, que é de difícil determinação experimental. Utilizamos um modelo simplificado, no caso a equação do cabo passivo, que é uma equação diferencial parabólica linear, com termos de difusão e reação, não necessariamente homogênea. Consideramos que o termo de reação é dado por uma função que depende da variável espacial, e é desconhecido. Para determinar essa função utilizamos o método de Landweber não linear, que, a partir de um ponto inicial qualquer (num espaço de Hilbert), busca de forma iterativa aproximações para a função desconhecida. Cada passo deste algoritmo requer a resolução de duas equações diferenciais parciais parabólicas e uma integral, utilizamos o método de Diferenças Finitas para obter a solução aproximada das equações diferenciais parciais e o método de trapézio para obter a solução da integral, resultando um método computacionalmente bastante intensivo. Nesta dissertação descrevemos a motivação biológica do problema, bem como a base matemática do algoritmo, e testamos vários casos computacionais.

Esta tese apresenta o desenvolvimento de uma técnica de processamento numérico capaz de reconstruir o sinal da temperatura do processo original a partir do sinal distorcido, atrasado e ruidoso, medido através de uma sonda intrusiva. Uma técnica de regularização foi adotada para contornar o mau condicionamento do modelo numérico inverso da equação de transdução para obter o sinal do processo, que considera o acúmulo térmico e as transferências de calor convectivo e radiativo entre o meio e o sensor térmico. O método dos mínimos quadrados simplificado foi implementado como técnica de regularização, por ser um método rápido e possuir um código computacional pequeno, permitindo, obter os dados em tempo real e desenvolver um sensor térmico inteligente. Testes numéricos demonstraram as discrepâncias introduzidas pela inércia térmica, convecção e radiação, bem como a extrema sensibilidade da presença de ruídos quando o problema inverso é resolvido. Testes experimentais foram conduzidos para validar o algoritmo de reconstrução sob condições práticas com sinais obtidos por um termopar encapsulado / This thesis presents the development of a numerical processing technique capable of reconstructing the original process temperature signal from distorted, late and noisy measured signal obtained from an intrusive probe. A regularization technique was adopted to palliate the ill conditioning of the inverse numerical model of the transduction equation to obtain the process signal, which considers thermal accumulation and convective and radiative heat transfers between the medium and the thermal sensor. The simplified least square method was implemented as the regularization technique, because it is a fast method and results in a small computational code, which produces on-line information and allow the development an intelligent temperature sensor. Numerical tests demonstrated the discrepancies introduced by thermal inertia, convection and radiation, as well as the extreme sensitivities to the presence of noise when solving the inverse problem. Experimental tests were carried out to validate the reconstruction algorithm under realistic experimental conditions with the signals obtained by the sheathed thermocouple

Dans ce travail de thèse, une méthode d’identification inverse des propriétés mécaniques du matériau viscoélastique (module de cisaillement et facteur de perte) fonctionnalisé dans une structure sandwich à trois couches symétriques est proposée. L’objectif de ce travail est de pouvoir identifier les propriétés mécaniques in situ. A travers un modèle éléments finis basé sur le modèle Zig-Zag de Rao, notre modèle assure la détermination des paramètres modaux de la poutre sandwich. L'approche inverse consiste en une procédure itérative qui détermine les formes de mode étant donné les paramètres matériaux puis calcule les propriétés viscoélastiques à partir des modes en utilisant un quotient de Rayleigh jusqu'à ce que la convergence sur les propriétés du matériau soit satisfaite. Les paramètres d’entrée du modèle inverse sont les fréquences de résonance et facteurs de perte de la poutre sandwich obtenues expérimentalement. En conséquence, la dépendance en fréquence des propriétés du matériau viscoélastique de la poutre sandwich est déterminée par une démarche automatisée. La méthode a été comparée avec succès aux formules de Ross-Kerwin-Ungar ; à une approche d'optimisation standard et à la littérature. A partir des résultats, nous avons pu déduire les lois de comportement du cœur viscoélastique suivant des modèles rhéologique tels que le modèle de Maxwell généralisé, d’ADF, de GHM et du Zéner fractionnaire. Ce dispositif expérimental couplé à la méthode d’indentification a permis l’investigation des paramètres modaux de la poutre à différentes températures pour étudier l’effet de la température sur les lois rhéologiques. Pour étudier la robustesse de notre méthode, nous avons procédé des essais de reproductibilité, de reproductibilité sur une population d’échantillon. L’efficacité de notre méthode étant été prouvée, une étude de sensibilité a été menée sur les caractéristiques géométriques de notre structure et les paramètres d’entrée. Les résultats obtenus montrent le fort impact de certains paramètres sur l’identification / In this work, a inverse identification method of the mechanical properties of the viscoelastic material (shear modulus and loss factor) functionalized in a sandwich structure with three symmetrical layers is proposed. The objective of this work is to be able to identify the mechanical properties in situ. Through a finite element model based on Rao's Zig-Zag model, our model ensures the modal parameter determination of the sandwich beam. The inverse approach consists of an iterative procedure that determines the mode shapes given the material parameters and then calculates the viscoelastic properties from the modes using a Rayleigh quotient until convergence on the properties of the material is satisfied. The input parameters of the inverse model are the resonance frequencies and loss factors of the sandwich beam obtained experimentally. As a result, the frequency dependence of the viscoelastic properties of the sandwich beam is determined by an automated way. The method has been successfully compared to Ross-Kerwin-Ungar formulas; a standard optimization approach and the literature. From the results, we have been able to deduce the constitutive laws of the viscoelastic heart according to rheological models such as the generalized Maxwell model, ADF, GHM and fractional Zener. This experimental device coupled to the method of identification allowed the investigation of modal parameters of the beam at different temperatures to study the effect of the temperature on the rheological laws. To study the robustness of our method, we carried out tests repeatability, reproducibility on a sample population. Since the effectiveness of our method has been proven, a sensitivity study has been carried out on the geometrical characteristics of our structure and the input parameters. The results obtained show the strong impact of certain parameters on identification

La magnéto-encéphalographie (MEG) mesure l´activité cérébrale avec un excellent décours temporel mais sa localisation sur la surface corticale souffre d´une mauvaise résolution spatiale. Le problème inverse MEG est dit mal-posé et doit de ce fait être régularisé. La parcellisation du cortex en régions de spécificité fonctionnelle proche constitue une régularisation spatiale pertinente du problème inverse MEG. Nous proposons une méthode de parcellisation du cortex entier à partir de la connectivité anatomique cartographiée par imagerie de diffusion. Au sein de chaque aire d´une préparcellisation, la matrice de corrélation entre les profils de connectivité des sources est partitionnée. La parcellisation obtenue est alors mise à jour en testant la similarité des données de diffusion de part et d´autre des frontières de la préparcellisation. C´est à partir de ce résultat que nous contraignons spatialement le problème inverse MEG. Dans ce contexte, deux méthodes sont développées. La première consiste à partitionner l´espace des sources au regard de la parcellisation. L´activité corticale est alors obtenue sur un ensemble de parcelles. Afin de ne pas forcer les sources à avoir exactement la même intensité au sein d´une parcelle, nous développons une méthode alternative introduisant un nouveau terme de régularisation qui, lorsqu´il est minimisé, tend à ce que les sources d´une même parcelle aient des valeurs de reconstruction proches. Nos méthodes de reconstruction sont testées et validées sur des données simulées et réelles. Une application clinique dans le cadre du traitement de données de sujets épileptiques est également réalisée. / Magnetoencephalography (MEG) is a functional non-invasive modality which provides information on the temporal succession of cognitive processes with an excellent time resolution. Unfortunately, spatial resolution is limited due to the ill-posed nature of the MEG inverse problem for estimating source currents from the electromagnetic measurement. Cortex parcellation into regions sharing functional features constitutes a relevant spatial regularization. We propose a whole cortex parcellation method based on the anatomical connectivity mapped by diffusion MRI. Inside areas of a preparcellation, the correlation matrix between connectivity profiles is clustered. The cortex parcellation is then updated testing the similarity of diffusion data on both sides of pre-parcellation boundaries. MEG inverse problem is constrained from this result. Two methods have been developed. The first one is based on the subdivision of source space regarding the parcellation. The cortical activity is obtained on a set of parcels and its analysis is simplified. Not to force sources to have exactly the same value inside a cortical area, we develop an alternative method. We introduce a new regularization term in the MEG inverse problem which constrain sources in a same region to have close values. Our methods are applied on simulated and real subjects. Clinical application is also performed on epileptic data. Each contribution takes part of a pipeline whose each step is detailed to make our works reproducible.

Em radioterapia, o espectro de energia é o componente mais importante dos feixes de elétrons. Espectros de energia de elétrons são relevântes para o cálculo acurado da dose, aplicações do sistema de planejamento e simulações realistas. Reconstrução inversa consiste na derivação do espectro de energia de elétrons a partir de curvas de porcentagem de dose em profundidade utilizando um apropiado modelo matemático. Reconstrução inversa é considerada a melhor dentre muitas abordagens porque: i) não requer nenhum equipamento suplementar ou do conhecimento detalhado da geometria e composição do cabeçote do acelerador; ii) equipamentos para a medição de curvas de porcentagem de dose em profundidade estão disponíveis em qualquer clínica e iii) é computacionalmente rápida. Neste trabalho, usou-se o método de reconstrução inversa baseado na sinergia recozimento simulado generalizado-regularização de Tikhonov. A validação da reconstrução foi realizada através do índice gama sob critérios clínicos de aceitação restritivos. Resultados mostraram que os espectros de energia reconstruídos reproduzem com precisão a porcentagem de dose em profundidade clínica bem como valores de dose fora do eixo central. Assim, concluí-se que o método empregado é ecaz para reconstruir espectros de energia que representam efetivamente espectros de energia do acelerador que atingem na supercie do fantoma. Consequentemente, sob certos limites, eles poderiam auxiliar em simulações realistas do tratamento. / In radiotherapy, energy spectrum is the most critical component of any electron beam. Knowledge of energy spectrum is important for accurate dose calculation, treatment planning applications and realistic simulations. Inverse reconstruction derives energy spectrum from the measured percentage depth dose using an appropriate mathematical model. There are several advantages to using inverse reconstruction: i) it does not require any supplementary equipment or detailed knowledge of the geometry head and composition; ii) the equipment for measurement of the percentage depth dose is standard and already available in any clinic and iii) it is computationally fast. In this work, we used the inverse reconstruction method based on the synergy simulated annealing generalized-Tikhonov regularization. Validation of inverse reconstruction was done by comparing the measured and reconstructed percentage depth dose via the gamma index. Results show the reconstructed electron energy spectra accurately reproduce the clinical dose percentage as well as o-axis dose values. Therefore, it was concluded that the method employed is eective to reconstruct energy spectra that eectively represent accelerator energy spectra reaching the phantom surface. Consequently, under certain limits, they could aid in realistic simulations of treatment.

Ce travail de thèse se situe à l’interface de deux domaines : la culture cellulaire en trois dimensions et l’imagerie sans lentille.Fournissant un protocole de culture cellulaire plus réaliste sur le plan physiologique, le passage des cultures monocouches (2D) à des cultures tridimensionnelles (3D) - via l’utilisation de gels extracellulaires dans lesquels les cellules peuvent se développer dans les trois dimensions - permet de faire de grandes avancées dans de nombreux domaines en biologie tels que l’organogénèse, l’oncologie et la médecine régénérative. Ces nouveaux objets à étudier crée un besoin en matière d’imagerie 3D.De son côté, l’imagerie sans lentille 2D fournit un moyen robuste, peu cher, sans marquage et non toxique, d’étudier les cultures cellulaires en deux dimensions sur de grandes échelles et sur de longues périodes. Ce type de microscopie enregistre l’image des interférences produites par l’échantillon biologique traversé par une lumière cohérente. Connaissant la physique de la propagation de la lumière, ces hologrammes sont rétro-propagés numériquement pour reconstruire l’objet recherché. L’algorithme de reconstruction remplace les lentilles absentes dans le rôle de la formation de l’image.Le but de cette thèse est de montrer la possibilité d’adapter cette technologie sans lentille à l’imagerie des cultures cellulaires en 3D. De nouveaux prototypes de microscopes sans lentille sont conçus en parallèle du développement d’algorithmes de reconstructions tomographiques dédiés.Concernant les prototypes, plusieurs solutions sont testées pour converger vers un schéma alliant deux conditions. La première est le choix de la simplicité d’utilisation avec une culture cellulaire en boîte de Petri standard et ne nécessitant aucune préparation spécifique ou aucun changement de contenant. Cette condition entraînant de fortes contraintes géométriques sur l’architecture, la deuxième est de trouver la meilleure couverture angulaire possible des angles d’éclairage. Enfin, une version adaptée aux conditions en incubateur est développée et testée avec succès.Concernant les algorithmes, quatre types de solutions sont proposés, basées sur le théorème de diffraction de Fourier classiquement utilisé en tomographie diffractive optique. Toutes cherchent à corriger deux problèmes inhérents au microscope sans lentille : l’absence de l’information de phase, le capteur n’étant sensible qu’à l’intensité de l’onde reçue, et la couverture angulaire limitée. Le premier algorithme se limite à remplacer la phase inconnue par celle d’une onde incidente plane. Rapide, cette méthode est néanmoins source de nombreux artefacts. La deuxième solution, en approximant l’objet 3D inconnu par un plan moyen, utilise les outils de la microscopie sans lentille 2D pour retrouver cette phase manquante via une approche inverse. La troisième solution consiste à implémenter une approche inverse régularisée sur l’objet 3D à reconstruire. C’est la méthode la plus efficace pour compenser les deux problèmes mentionnés, mais elle est très lente. La quatrième et dernière solution est basée sur un algorithme de type Gerchberg-Saxton modifié avec une étape de régularisation sur l’objet.Toutes ces méthodes sont comparées et testées avec succès sur des simulations numériques et des données expérimentales. Des comparaisons avec des acquisitions au microscope classique montrent la validité des reconstructions en matière de tailles et de formes des objets reconstruits ainsi que la précision de leur positionnement tridimensionnel. Elles permettent de reconstruire des volumes de plusieurs dizaines de millimètres cubes de cultures cellulaires 3D, inaccessibles en microscopie standard.Par ailleurs, les données spatio-temporelles obtenues avec succès en incubateur montrent aussi la pertinence de ce type d’imagerie en mettant en évidence des interactions dynamiques sur de grandes échelles des cellules entres elles ainsi qu’avec leur environnement tridimensionnel. / This PhD work is at the interface of two fields: 3D cell culture and lens-free imaging.Providing a more realistic cell culture protocol on the physiological level, switching from single-layer (2D) cultures to three-dimensional (3D) cultures - via the use of extracellular gel in which cells can grow in three dimensions - is at the origin of several breakthroughs in several fields such as developmental biology, oncology and regenerative medicine. The study of these new 3D structures creates a need in terms of 3D imaging.On another side, 2D lens-free imaging provides a robust, inexpensive, non-labeling and non-toxic tool to study cell cultures in two dimensions over large scales and over long periods of time. This type of microscopy records the interferences produced by a coherent light scattered by the biological sample. Knowing the physics of the light propagation, these holograms are retro-propagated numerically to reconstruct the unknown object. The reconstruction algorithm replaces the absent lenses in the role of image formation.The aim of this PhD is to show the possibility of adapting this lens-free technology for imaging 3D cell culture. New lens-free microscopes are designed and built along with the development of dedicated tomographic reconstruction algorithms.Concerning the prototypes, several solutions are tested to finally converge to a scheme combining two conditions. The first requirement is the choice of simplicity of use with a cell culture in standard Petri dish and requiring no specific preparation or change of container. The second condition is to find the best possible angular coverage of lighting angles in regards of the geometric constraint imposed by the first requirement. Finally, an incubator-proof version is successfully built and tested.Regarding the algorithms, four major types of solutions are implemented, all based on the Fourier diffraction theorem, conventionally used in optical diffractive tomography. All methods aim to correct two inherent problems of a lens-free microscope: the absence of phase information, the sensor being sensitive only to the intensity of the incident wave, and the limited angular coverage. The first algorithm simply replaces the unknown phase with that of an incident plane wave. However, this method is fast but it is the source of many artifacts. The second solution tries to estimate the missing phase by approximating the unknown object by an average plane and uses the tools of the 2D lens-free microscopy to recover the missing phase in an inverse problem approach. The third solution consists in implementing a regularized inverse problem approach on the 3D object to reconstruct. This is the most effective method to deal with the two problems mentioned above but it is very slow. The fourth and last solution is based on a modified Gerchberg-Saxton algorithm with a regularization step on the object.All these methods are compared and tested successfully on numerical simulations and experimental data. Comparisons with conventional microscope acquisitions show the validity of the reconstructions in terms of shape and positioning of the retrieved objects as well as the accuracy of their three-dimensional positioning. Biological samples are reconstructed with volumes of several tens of cubic millimeters, inaccessible in standard microscopy.Moreover, 3D time-lapse data successfully obtained in incubators show the relevance of this type of imaging by highlighting large-scale interactions between cells or between cells and their three-dimensional environment.

<p> The inverse heat conduction problem also frequently referred as the sideways heat equation, in short SHE, is considered as a mathematical model for a real application, where it is desirable for someone to determine the temperature on the surface of a body. Since the surface itself is inaccessible for measurements, one is restricted to use temperature data from the interior measurements. From a  mathematical point of view, the entire situation leads to a non-characteristic Cauchy problem, where by using recorded temperature one can solve a well-posed nonlinear problem in the finite region for computing heat flux, and consequently obtain the Cauchy data [u, u_x]. Further by using these data and by performing an appropriate method, e.g. a space marching method, one can eventually achieve the desired temperature at x = 0.</p><p>The problem is severely ill-posed in the sense that the solution does not depend continuously on the data. The problem solved by two different methods, and for both cases we stabilize the computations by replacing the time derivative in the heat equation by a bounded operator. The first one, a spectral method based on finite Fourier space is illustrated to supply an analytical approach for approximating the time derivative. In order to get a better accuracy in the numerical computation, we use cubic spline function for approximating the time derivative in the least squares sense.</p><p>The inverse problem we want to solve, by using Cauchy data, is a nonlinear heat conduction problem in one space dimension. Since the temperature data u = g(t) is recorded, e.g. by a thermocouple, it usually contains some perturbation in the data. Thus the solution can be severely ill-posed if the Cauchy data become very noisy. Two experiments are presented to test the proposed approach.</p>