Morphology-driven automatic segmentation of MR images of the neonatal brain

► We propose a fully automatic method for the segmentation of neonatal brain MRI. ► It segments brain parts and tissues essential for evaluating early brain development. ► It uses high-level brain morphology priors, and needs no atlas or manual intervention. ► Priors concern tissue structure, connectivity and relative positions. ► Results are visually accurate and show high overlap with expert manual segmentation.