Abstract : In the present study, we propose an automatic framework for obtaining an accurate representation of the in-vivo fetal brain for the quantification of cerebral volume and cortical surface. One of the biggest challenges in performing non-sedated fetal MRI studies is fetal motion, resulting in motion artifact an poor image quality. To date, most conventional fetal MRI acquisitions will have a 1x1x2mm3 resolution. The challenge is then to recover a high image quality from such volumes, in order to reliably characterize cortical development during intra-uterine life. Typically, three fetal scans are acquired, one in each direction, providing high in-plane resolution slices in each direction. The problem is then to create a single, isotropic high-resolution volume. This reconstruction is detailed herein, as well as results based on synthetic and real fetal data. An example of fetal cortical surface rendering is also presented.
