Résumé
Over these last years, new computerized methods have been proposed to build virtual endocasts based on 3D CT images of extant or fossil skulls. This allows analyzing large samples automatically and then obtaining significant statistical results about the shape of the endocast and its correlation relative to other anatomical structures. We assessed two new algorithms to segment and process virtual endocasts. The first tool allows delineating the boundary of the 3D endocranial surface without any manual interaction, even in presence of large holes (such as the foramen magnum). We compared these virtual endocasts with the corresponding plaster endocasts manually casted by one the authors. The second algorithm allows the automated measurement of 3D asymmetry on the virtual endocasts, and the computation of a mean 3D endocranial shape over several individuals on which the mean population asymmetry can be assessed. To show the potential of such automatic tools, we propose to study the relationship between the shape of the endocast and the external shape of the skull. For this purpose, we selected about twenty dry skulls in the Morton collection. These skulls were selected from populational samples which have been considered in the past to evince pronounced exo-cranial secondary-developing super-structures. These samples were CT-scanned and we automatically segmented both the external surface of the skull and the endocast. We present the first results of our analysis by focusing on the exocranial versus endocranial differences at the level of the neurocranium of each individual. We also compare asymmetry patterns between individuals.