Best Practices for Artificial Intelligence and Machine Learning for Computer-Aided Diagnosis in Medical Imaging

Abstract

The American Association of Physicists in Medicine (AAPM) Task Group (TG) 273 has been charged with developing recommendations on best practices for the development and performance assessment of computer-aided decision support systems. The TG report [1] addresses broad issues common to the development of most, if not all, CAD-AI applications and their translation from the bench to the clinic. The goal was to bring attention to issues such as proper data collection and training and validation methods for ML algorithms, aiming to improve generalizability and reliability and thus accelerate the adoption of CAD-AI systems for clinical decision support. The report focuses on several developmental stages of CAD-AI: data collection, reference standards, model development, performance assessment, and translation to the clinic.