An Editor for the Acquisition of Medical Expert Knowledge: Function and Use shown by an Automatic Knowledge-based System for the Analysis of Heart Scintigrams

Abstract

The last 20 years have shown a rapid development in building expert or knowledge-based systems. Those systems are one of the major objectives in research concerning artificial intelligence. The areas of application vary from theorem proving, industry automation, control and, especially, medicine. Expert systems like MYCIN [BUC84], INTERNIST [MIL82], CASNET [WEI81] or PATREC [MIT84] are well-known from literature. Perhaps in near future, they will do their work in laboratories and clinical environment in everyday use, that is they will guide a physician interactively to a decision concerning the special application problem. Another aspect is to develop knowledge-based systems solving difficult problems without any user interactions required. This method is favourable if input data are produced by a camera, for instance, which could store them in the computer automatically. Applications would be in ultrasound, radiology, CT, MRI and, especially, nuclear medicine involving a lot of medical imaging tasks. Up to now, interpretation of the resulting images is usually performed by visual inspection. This method is rather limited in many respects (e.g. estimation of greyscale thresholds, dependence on image displaying devices etc. affect a practical interpretation of those data) and necessitates the repetitive recording of identical phenomena from different points of view.