Toward an automated analysis system for nuclear magnetic resonance imaging. II. Initial segmentation algorithm.

Abstract

Image segmentation algorithms based on hierarchical clustering have been developed for analysis of T1 and T2 nuclear magnetic resonance images. Application of these algorithms to simultaneous T1-T2 images of healthy volunteers extracted fundamental tissue types in the brain. These algorithms also were used both to identify the extent of the region of involvement of a subject with a history of a grade 3 astrocytoma of the right frontal lobe of the brain, and to characterize the tissue within the region of involvement. These results suggest that a simple segmentation algorithm can produce reasonable clustering of tissue types within the brain.