AI and chest X-rays

The use of artificial intelligence (AI) in healthcare has risen dramatically since its first description in the 1950s, enabled by the development of deep learning¹. Research surrounding these systems is focused largely on deep learning algorithms (DLAs). The aim of this literature review is to collate and provide context to key research surrounding multiple abnormality detection in chest X-rays (CXRs) using AI software. Through determining where knowledge is still lacking, future work can be influenced, allowing for developments to meet relevant standards and direct future policy-making and clinical practice.

The papers selected for this review were published between the years of 2017 and 2022 and focused on classification of CXRs using AI software. The analysis was thematic², with the overall value of the papers considered through exploration of the methods, results and limitations. The success demonstrated within these papers supports the...