Out-of-distribution Detection in Medical Image Analysis: A survey

TL;DR

This survey reviews recent advances in out-of-distribution detection methods for medical image analysis, highlighting challenges, evaluation metrics, and future research directions to improve trustworthy AI in clinical settings.

Contribution

It provides a systematic categorization and review of OOD detection techniques in medical imaging, addressing factors causing distributional shifts and evaluation protocols.

Findings

Various factors cause distributional shifts in clinical scenarios

A framework categorizes existing OOD detection solutions

Discussion on evaluation metrics and future research directions

Abstract

Computer-aided diagnostics has benefited from the development of deep learning-based computer vision techniques in these years. Traditional supervised deep learning methods assume that the test sample is drawn from the identical distribution as the training data. However, it is possible to encounter out-of-distribution samples in real-world clinical scenarios, which may cause silent failure in deep learning-based medical image analysis tasks. Recently, research has explored various out-of-distribution (OOD) detection situations and techniques to enable a trustworthy medical AI system. In this survey, we systematically review the recent advances in OOD detection in medical image analysis. We first explore several factors that may cause a distributional shift when using a deep-learning-based model in clinic scenarios, with three different types of distributional shift well defined on top of these factors. Then a framework is suggested to categorize and feature existing solutions, while the previous studies are reviewed based on the methodology taxonomy. Our discussion also includes evaluation protocols and metrics, as well as the challenge and a research direction lack of exploration.