Multispectral Analysis Of Magnetic Resonance Images: A Comparison Between Supervised And Unsupervised Classification Techniques
Published 1990 · Computer Science
With six or more tissue parameters MR imaging has substantial theoretical potential for tissue discrimination in different organs. Accurate tissue discrimination may require dependence on computer-based image analysis techniques to extract tissue-specific information. Our approach to these problems has been through pattern classification techniques (Alaux 1988, 1989). Our routine analysis has included: principal components analysis, supervised Bayesian classification, and clustering techniques (minimum distance and dynamic clustering).