[en] This paper reports a recent study on applying a knowledge-based system approach as a new attempt to solve the problem of chromosome classification. A theoretical framework of an expert image analysis system is proposed, based on such a study. In this scheme, chromosome classification can be carried out under a hypothesize-and-verify paradigm, by integrating a rule-based component, in which the expertise of chromosome karyotyping is formulated with an existing image analysis system which uses conventional pattern recognition techniques. Results from the existing system can be used to bring in hypotheses, and with the rule-based verification and modification procedures, improvement of the classification performance can be excepted