Soft-Tissue Segmentation in Forensic Applications

In this paper a new method for extracting a soft facial template from T1 weighted magnetic resonance (MR) images for craniofacial reconstruction is proposed. The reconstruction of the face of dead individuals based on the shape of their skull is for interest in forensic and archaeology as well as in anthropology. Several computational methods have been developed to simplify facial reconstructions. The aim of the presented multimodality approach is to extract a facial template from MR images and register this template on the computed tomography (CT) of the skull of a dead individual. In the following first results of an automatic segmentation strategy to obtain the desired facial template are presented. The approach is based on active contours, introduced by M. Kass et al. [1], which are well known from image processing applications to locate boundary objects. Additionally to the classical formulation their gradient vector flow based version is presented. The resulting flow field leads to a contour, which is capable of entering concave regions. All current results are presented in 2D.