AI-based Retrospective Study for Revealing Diagnostic Errors in Chest X-ray Screening

Abstract

In this paper, we explore the ability of an AI-based computer-aided diagnostic system (CAD) to help to reveal the early signs of probable lung diseases in X-ray images. We use a large screening database which contains natively-digital X-ray images acquired between 2001 and 2014 along with the corresponding diagnostic reports provided by the radiologists. We apply a Deep Learning-based CAD system to the cases from the database which were labeled by the radiologist as a norm and compare the CAD prediction results to the radiologists’ diagnostic reports. Our experiments demonstrate the ability of an automated AI-based CAD to reveal discrepancies between the diagnostic reports and the actual state of lungs as conveyed by the X-Ray image. Additionally, in a number of cases the Deep Learning algorithm was able to detect early signs of lung diseases which progressed later according to the patient anamnesis.