Interpretable artificial intelligence systems in medical imaging: review and theoretical framework

Abstract

The development of Interpretable Artificial Intelligence (AI) has drawn substantial attention on the effect of AI on augmenting human decision-making. In this paper, we review the literature on medical imaging to develop a framework of Interpretable AI systems in enabling the diagnostic process. We identify three components as constituting Interpretable AI systems, namely, human agents, data, machine learning (ML) models, and discuss their classifications and dimensions. Using the workflow process of AI augmented breast screening in the UK as an example, we identify the possible tensions that may emerge as human agents work with ML models and data. We discuss how these tensions may impact the performance of Interpretable AI systems in the diagnostic process and conclude with implications for further research.

Ähnliche Arbeiten