MAGNETIC-RESONANCE IMAGE SEGMENTATION USING PATTERN-RECOGNITION, AND APPLIED TO IMAGE REGISTRATION AND QUANTITATION

This review highlights various magnetic resonance image (MRI) segmenta tion algorithms that employ pattern recognition. The procedures are gr ouped into two categories: low- to intermediate-level, and high-level image processing. The former consists of grey level histogram analysis , texture definition, edge identification, region growing, and contour following. The roles of significant prior knowledge, neural networks and cluster analysis are examined by producing objective identificatio n of anatomical structures. The application of the segmented anatomica l structures in image registration, to monitor the disease progression or growth of anatomy in normal volunteers and patients, is highlighte d. The use of the segmented anatomy in measuring volumes of structures in normals and patients is also examined.