Expert systems in histopathology. III. Representation of knowledge as "structured objects".

In designing expert systems for histopathology, the representation of knowledge as "structured objects" may offer advantages over the use of a set of rules. A structured-object representation using straightforward, descriptive, declarative statements can allow consideration of the relationships between the facts and constraining conditions of a domain in a richer, more complex and more complete manner. The use of associative (semantic) networks and frames to organize histopathologic knowledge as structured objects is discussed and demonstrated with examples that consider the complex tissue structures found in the kidney and thyroid. As an example, the design and function of a structured object-based expert system for follicular thyroid aspirates is discussed, including a consideration of the types of information that such a system would require. Such a system utilizes model-based reasoning rather than the rule-based reasoning of rule-based expert systems.