Volumetric Assessment of Brain MRIs of Fetuses with Spina Bifida

Motivation: Spina bifida occurs during the first gestational month and causes lifelong disabilities. If fetuses with spina bifida are left undiagnosed or untreated, spinal cord defects are translated into brain abnormalities. Goal(s): Our goal is to annotate brain regions in fetal MRIs to study typical and atypical fetal brain development in spina bifida. Approach: We developed a deep learning fetal brain MRI segmentation method and modeled growth to statistically compare brain volumes of normal and pathological cohorts. Results: Our segmentation method reliably annotates fetal brain MRIs. We observed significant increase in the ventricles and significant reduction of the cerebellum in fetuses with spina bifida. Impact: Fetal brain MRI segmentation with our segmentation model enables precise delineation of brain tissues and anatomical structures, allowing early detection of aberrant brain development due to congenital defects such as spina bifida.