Abstract : A definite diagnosis of Parkinson's disease (PD) and multiple system atrophy (MSA) can only be confirmed neuropathologically. The clinical differentiation of the parkinsonian variant of multiple system atrophy (MSA-P) from PD is challenging, especially during the early stages of the disease. Early differentiation of these diseases is particularly important because the disorders differ in terms of progression, prognosis, and treatment responses. The aim of the first part of the study was to evaluate in vivo changes in the brain's macro- and microstructure in MSA-P and in PD and to characterize the cerebral anatomical differences between the two conditions. We used a combination of voxel-based morphometry (VBM) and whole-brain, voxel-based diffusion tensor imaging analysis (VB-DTI). In MSA-P patients, VBM analysis revealed a lower density of grey matter (GM) in a motor related circuit (especially in the left primary motor cortex, PMC), relative to PD patients, and in the left supplementary motor area (SMA), relative to controls). Diffusion tensor imaging analysis revealed lower fractional anisotropy (FA) values in the left PMC and the right cerebellum in MSA-P patients, compared with controls. Using a volumetric diffusion technique, our study revealed selective tissue degeneration in motor circuits, regardless of the volume loss detected in VBM and in agreement with pathology reports and clinical motor characteristics. Our findings suggest that MSA-P is characterized by both macro- and microstructural changes in the sensorimotor circuit. The aim of the second part of the study was to evaluate brain connectivity especially cortico putaminal connectivity using probabilistic tractography. We assessed volumetric and microstructural changes that occur within each of these subregions and try to establish the potential value of these changes in differential diagnosis.DTI and T1-weighted images were obtained using 1.5 T MRI. Putamen was manually segmented. The cortex was segmented using Freesurfer software and cortical regions were classified in three functional systems: motor, associative and limbic. Then, we calculated the connection probability between putamen and ipsilateral cortical target. Volumes and DTI parameters (fractional anisotropy FA, mean diffusivity MD) of the resulting DTI-based parcellations of the putamen were compared between groups. Comparisons between groups were carried out using bivariate non parametrics tests. Putamen microstructural changes were present in the two variants of MSA according to anatomopathological knowledge. Loss of motor connectivity in MSA-P patients can be explained partially by important volume loss of putamen. Statistical multivariate model combining few clinical criteria and data obtained by MRI-based parcellation allows discriminating MSA-P from MSA-C from PD patients and controls in more than 75% of cases.The third part's aim was to study the cognitive profile of MSA patients compared to PD patients and to evaluate the cognitive clinical correlations with VBM and VB-DTI brain MRI data. The contribution of standard neuropsychological examination to the differential diagnosis of both syndromes remains still limited. Our study revealed the main involvement of motor cortex in cognitive functions. From our knowledge, there is no study of cognitive correlations and DTI parameters in PD or in MSA.