Matching the cortex across subjects is a central problem in neuroimaging because it is the condition to perform anatomical or functional studies across large groups of subjects. Nevertheless, the large inter-individual variability of the cortex makes this problem ill defined. The standard solution is to register any individual to a standard average template using smooth deformations and using a similarity function that depends on geometrical (curvature) or intensity information. Yet, nobody knows to what extent the matching of two different brains with a continuous deformation makes sense. In fact, the problem of inter-subject comparison is being treated as a registration issue when it should be treated as a localization one. The real question that should asked when looking at a specific location is "where am I ?" Answering this question implicitely solves the problem of correspondances across subjects. In this talk I will present some of our recent work on cortical localization and inter-subject matching of cortical surfaces. We use a model of cortical organization developed in collaboration with Prof. J. Régis, neurosurgeon at La Timone Hospital, Marseille. This model provides an abstract description of how the cortex is organized in terms of macro-anatomy (folds). We also provide an implementation on any given cortical surface under the form of a 2D coordinate system constrained by specific anatomical features. I will present the anatomical theories behind the model, the technical implementation we proposed using conformal mappings to a rectangular domain, as well as results on real data and possible applications
******** CSS e script comuni siti DOL - frase 9957 ********p>