KNOWLEDGE-BASED CLASSIFICATION AND TISSUE LABELING OF MR-IMAGES OF HUMAN BRAIN

This paper presents a knowledge-based approach to automatic classifica tion and tissue labeling of 2-D magnetic resonance (MR) images of the human brain. The system consists of two components: an unsupervised cl ustering algorithm and an expert system. MR brain data is initially se gmented by the unsupervised algorithm, then the expert system locates a landmark tissue or cluster and analyzes it by matching it with a mod el or searching in it for an expected feature. The landmark tissue loc ation and its analysis are repeated until a tumor is found or all tiss ues are labeled. The knowledge base contains information on cluster di stribution in feature space and tissue models. Since tissue shapes are irregular, their models and matching are specially designed: 1) quali tative tissue models are defined for brain tissues such as white matte r; 2) default reasoning is used to match a model with an MR image; tha t is, if there is no mismatch between a model and an image, they are t aken as matched. The system has been tested with fifty-three slices of MR images acquired at different times by two different scanners. It a ccurately identifies abnormal slices and provides a partial labeling o f the tissues. It provides an accurate complete labeling of all normal tissues in the absence of large amounts of data non-uniformity, as ve rified by radiologists. Thus the system can be used to provide automat ic screening of slices for abnormality. It also provides a first step toward the complete description of abnormal images for use in automati c tumor volume determination.