Symbolic reasoning about myocardial scintigrams in PROLOG

Area of Science:

• Artificial Intelligence
• Medical Imaging
• Computer Science

Background:

• PROLOG (PROgramming in LOGic) is a declarative programming language central to the Japanese fifth-generation computer project.
• Symbolic image processing is an area where PROLOG shows potential utility.
• Accurate prediction of coronary anatomy from myocardial scintigrams is clinically significant.

Purpose of the Study:

• To evaluate PROLOG as a tool for symbolic image processing in the context of medical diagnosis.
• To develop a system for predicting coronary anatomy using myocardial scintigrams.
• To incorporate methods for handling uncertainty in the diagnostic process.

Main Methods:

• Utilizing PROLOG for symbolic image processing tasks.
• Applying a combination of fuzzy-set theoretic and probabilistic reasoning to manage uncertainty.
• Developing heuristic classification rules based on clinical expertise.
• Training and validating the system on a dataset of 247 myocardial scintigrams and corresponding coronary angiograms.

Main Results:

• Demonstrated the suitability of PROLOG for symbolic image processing in a medical application.
• Successfully predicted coronary anatomy from myocardial scintigrams.
• Effectively managed uncertainty using integrated fuzzy-set and probabilistic approaches.

Conclusions:

• PROLOG is a viable tool for symbolic image processing in medical applications like coronary anatomy prediction.
• The combined fuzzy-set and probabilistic reasoning effectively addresses uncertainty in diagnostic tasks.
• Heuristic rules derived from clinical experience and data enhance diagnostic accuracy.