Morphométrie cérébrale et variabilité corticale – BrainMorph

Brain morphometry has been an emerging field in the past few years, since the variety of brain data that can be acquired in vivo (in particular with magnetic resonance imaging) has expanded and the computer-assisted techniques necessary to analyze those data have greatly improved. It is now acknowledged that morphometric techniques can be used both in the early diagnosis of brain pathology, a subject of great clinical importance, as well as for the more theoretical understanding of normal and abnormal brain variability. One recent approach, which consists of analyzing data on the cortical surface (bi-dimensional) instead of the acquisition volume, has raised particular interest. This change of data definition domain (from a 3D Euclidean domain to a 2D Riemanian manifold) has many advantages, but existing volume-based techniques cannot be applied to surface-based analysis. It is therefore necessary to solve several problems: - how to match cortical surfaces derived from different subjects' ' which measurements and features should be extracted in order to evaluate cortical variability, to describe cortical organization, or to characterize pathology ' ' is there a reference space within which we can work at the population level, which would be essential for analyzing, localizing, and communicating information. In other words, can we define an atlas of the cortical surface' This project aims at addressing these issues and is organised around several aims and research lines: - improving our understanding and model of cortical organization, and its implementation, by bringing it to a finer level of description in order to reduce normal variability when performing the necessary inter-subject matching. ' proposing a set of surface-based morphometric tools, feature extraction tools, and pattern recognition techniques, to describe gyral patterns on the cortical surface. ' building an atlas of the cortical surface and a metric associated with it. This also implies developing methods for statistics on shape in order to describe and quantify cortical variability. ' applying those tools to a large database of normal and pathological subjects. ' providing all necessary software implementation to bring the methodological developments to a stage where they can be used by the neuroimaging community, via the BrainVisa software platform. The project brings together teams of methodologists, neuroscientists, clinicians, and MRI physicists in order to address these issues and to demonstrate the relevance of our answers for real clinical and theoretical applications. The originality lies in the model-driven description of the cortical surface and its application to brain morphometry. We aim to investigate cortical variability and to propose a methodology for surface-based morphometric techniques. We also aim to propose an atlas of the cortical surface and make it a reference for cortical localization. Finally, we want to bring these methods to a community of users, by putting a strong emphasis on the software implementation of the proposed techniques.