Le développement de méthodes alternatives à la culture in vivo pour l’évaluation de la toxicité des molécules chimiques s’est accéléré ces dernières années, l’objectif étant de limiter l’utilisation d’animaux. Préconisés par l’OCDE (Organisation de coopération et de développement économiques), ces modèles alternatifs visent à mimer les conditions physiologiques en employant des systèmes in vitro ou in silico. Parmi les différents systèmes développés, les biopuces microfluidiques ont prouvé leur contribution à l’amélioration des fonctions cellulaires, ce qui permet des études toxicologiques pertinentes. Les travaux de ce doctorat sont basés sur l’emploi de ces biopuces pour cultiver des hépatocytes (cellules du foie) et portent sur la mise au point d’une méthode d’analyse d’images issues de ces cultures sous microscope au cours du temps. L’acquisition d’images tout au long de l’expérience permet de suivre, après traitement, l’évolution et le comportement des cellules au contact de molécules chimiques et d’évaluer les réponses toxicologiques. Les premiers résultats de ces travaux ont permis l’amélioration du procédé de culture microfluidique adaptée au matériel d’acquisition d’images, la sélection de sondes fluorescentes, et le choix d’un algorithme de traitement des images sur CellProfiler. Cela nous a permis de quantifier et caractériser certaines fonctions biologiques au sein de la biopuce comme l’activité mitochondriale. Le potentiel de cet outil pour évaluer la toxicité de molécule a été testé grâce à l’emploi d’un toxique connu : la staurosporine. Les résultats obtenus ont révélé l’impact de la mise en culture en dynamique sur le comportement des hépatocytes, et la toxicité de la staurosporine visible en biopuce. / The development of alternative methods of in vivo cultures for the toxicological evaluation of chemical molecules has accelerated this last years, in order to limit the use of animals. Recommended by the OECD (Organisation for Economic Cooperation and Development), these alternative models are designed to mimic the physiological conditions using in vitro or in silico systems. Among the developed systems, microfluidic biochips have proven their contribution to the improvement of cellular functions, which allows relevant toxicological studies. This PhD thesis is based on the use of these biochips for hepatocytes culture and focus on the development of an analysis method for study these cultures under microscope over time using imaging. Image acquisition throughout the experiment enables to analyze, after image processing, the evolution and the behavior of cells in contact with chemical molecules and to evaluate toxicological responses. The first results of this work led to the optimization of the microfluidic cultures under the microscope used to get the image sequences, the selection of fluorescent probes and the development of an image processing system with CellProfiler. These works allowed the quantification and the characterization of some biological functions within the biochip such as the mitochondrial activity. Staurosporine, a well-known toxic, has been used to test the potential of this tool to evaluate the toxicity of molecules. The results showed the impact of dynamic culture on the hepatocytes behavior, and the staurosporine toxicity, in biochip cultures.

<p>Flow-based microfluidic biochips constitute an emerging technology for the automation of biochemical procedures. Recent advances in fabrication techniques have enabled the development of these devices. Increasing integration levels provide biochips with tremendous potential; a large number of bioassays, i.e., protocols for biochemistry, can be processed independently, simultaneously, and automatically on a coin-sized microfluidic platform. However, the increases in integration level introduce new challenges in the design optimization and the testing of these devices, which impede their further adoption and deployment.</p><p>This thesis is focused on enhancing the automated design and use of flow-based microfluidic biochips and on developing a set of solutions to facilitate the full exploitation of design complexities that are possible with current fabrication techniques. Four key research challenges are addressed in the thesis; these include design automation, wash optimization, testing, and defect diagnosis.</p><p>Despite the increase in the number of on-chip valves, designers are still using full-custom methodologies involving many manual steps to implement these chips. Since these chips can easily have thousands of valves, manual design procedure can be time-consuming and error-prone, and it can result in inefficient designs. This thesis presents the first problem formulation for automated control-layer design in flow-based microfluidic biochips and describes a systematic approach for solving this problem. Our goal is to find an efficient routing solution for control-layer design with a minimum number of control pins.</p><p>The problem of contamination removal in flow-based microfluidic biochips must also be addressed. Applications in biochemistry require high precision to avoid erroneous assay outcomes, and they are vulnerable to contamination between two fluidic flows with different biochemistries. This thesis proposes the first approach for automated wash optimization for contamination removal in flow-based microfluidic biochips. The proposed approach ensures effective cleaning and targets the generation of wash pathways to clean all contaminated microchannels with minimum execution time under physical constraints.</p><p>Another practical problem addressed in this thesis is the lack of test techniques for screening defective biochips before they are used for biochemical analysis. This thesis presents an efficient approach for automated testing of flow-based microfluidic biochips. The test technique is based on a behavioral abstraction of physical defects in microchannels and valves. The flow paths and flow control in the microfluidic device are modeled as a logic circuit composed of Boolean gates, which allows test generation to be carried out using standard automatic test-pattern generation tools. Based on the analysis of untestable faults in the logic-circuit model, we present a design-for-testability technique that can achieve 100\% fault coverage.</p><p>Finally, this thesis presents a technique for the automated diagnosis of leakage and blockage defects. The proposed method targets the identification of defect types and their locations based on test outcomes. It reduces the number of possible defect sites significantly while identifying their exact locations.</p><p>In summary, this thesis has led to a set of optimization and testing methods for flow-based microfluidic biochips. The results of this research are expected to not only shorten the product development cycle, but also accelerate the adoption and further development of this emerging technology by facilitating the full exploitation of design complexities that are possible with current fabrication techniques.</p> / Dissertation