Three dimensional image processing and an expert system for tissue characterization studies by NMR

This paper describes new three dimensional image processing methods and an expert system for tissue characterization studies by nuclear magnetic resonance (NMR). Proposed methods for plane reconstruction from a three dimensional NMR image (reconstruction of two dimensional arbitrary planes) and surface reconstruction can be available to structural analysis of biological tissues, detection of diseased regions and positioning for medical treatments. Arbitrary two dimensional plane image and surface image reconstructed in the rectangular and the polar coordinates provide useful structural information. On the other hand, the expert system for tissue characterization studies evaluates and classifies biological tissues in a extracted ROI based on results of numerical analysis of measured values, the spin density and the relaxation times (T1 and T2), and extracted features from two and three dimensional NMR images. The system is equipped with a knowledge base concerning these measured values in normal and pathological human, rat and rabbit tissues. The LISP language is used for evaluating the tissue characterization. The system is installed in a computer, IBM 3081 K, and can directly communicate with a processing system for three dimensional images and a picture archiving and communication system (PACS) in another computer, FACOM M380. (author)