Segmentation of myelinated white matter in pediatric brain magnetic resonance images

摘要

The automated tissue classification of pediatric brain magnetic resonance images, specially proper segmentation of myelinated white matter, is a highly challenging task. The proposed approach first extracts the brain tissue, followed by a precise delineation of the myelinated component based on Tsallis entropy segmentation. Unlike most of the currently available algorithms, the proposed technique is totally atlas-free. Qualitative validation shows that the obtained segmentation results correspond well to those of manual segmentation.