Information based approach of the phase contrast imaging by synchrotron radiation. Preclinical applications to brain imaging of the small animal

Virtual histology is a field of investigation with growing interest in the community of bio-image analysis. We focus on one of these techniques with the phase contrast tomography using synchrotron radiation. This technique allows us to visualize mice brains with no impact and with a spatial resolution of 8 μm isotropically, which is a resolution similar to the one obtained with classic optical histology but without damaging samples with specific dyeing. This thesis is organized along three main axes. The first one presents photonic instrumentation which gives us access to the phase information and the original setting of a reconstruction parameter for the acquisition of biological heterogeneous samples. A second axis shows several image processing developed in order to address different informational tasks such as visual optimization, cellular aggregates detection or fiber tractography. Eventually, a biomedical application of these process is proposed by addressing detection and quantification of iron oxide nanoparticles in an experimental model of stroke. (author)