Problems in applying expert system technology to radiographic image interpretation.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Radiology

Background:

• Expert systems offer potential for automating radiographic image interpretation.
• Challenges exist in applying AI to complex medical diagnostics.
• The Radiographic Image Interpretation System (RIIS) was developed to address these challenges.

Purpose of the Study:

• To evaluate the practical application and identify potential problems of an expert system for radiographic image interpretation.
• To assess the performance of the RIIS prototype with both experienced and inexperienced users.

Main Methods:

• Developed a prototype expert system, the Radiographic Image Interpretation System (RIIS), using Turbo Prolog on a microcomputer.
• Focused the system on a limited domain of focal bony lesions.
• Evaluated RIIS with 20 pathologically proven cases at the 1987 Radiological Society of North America (RSNA) meeting, comparing results from inexperienced and experienced users.

Main Results:

• When used by a musculoskeletal radiologist familiar with its operation, RIIS achieved correct diagnoses in the top five 80% of the time.
• During the RSNA meeting, with inexperienced users, RIIS selected the correct diagnosis in the top five only 22% of the time.

Conclusions:

• The prototype expert system (RIIS) demonstrates potential in radiographic image interpretation but faces significant usability challenges with inexperienced users.
• Further development is needed to improve the performance and accessibility of expert systems for clinical radiographic interpretation.