Automatic segmentation and labeling of human brain tissue from MR images

Abstract

This work presents a technique for automatic tissue labeling of 2-D magnetic resonance (MR) images of the human brain. This technique consists of two components: an unsupervised clustering algorithm and a knowledge-based technique. The knowledge-based technique contains information on the cluster distribution in feature space and tissue models. This approach also provides a first step toward classification of normal and abnormal images.