Figure 5.

Schematic flow of the proposed segmentation approach for myelin-like signals (MLS) on T₂-weighted neonatal brain images. Through image preprocessing, we achieve brain extraction and bias field correction, and obtain the binary masks of the deep brain region and brainstem for individual subjects. The automatic segmentations of the deep gray matter (DGM), obtained using the Statistical Parametric Mapping (SPM) software, contain the basal ganglia, thalami, hippocampi and amygdalae. We extract the masks for the deep brain region from the DGM segmentations via label propagation.