Diagnostic Accuracy of Artificial Intelligence-Based Chest X-Ray reading for screening of Tuberculosis

Abstract

Background: Tuberculosis remains a public health challenge in Nepal and ranks as the seventh leading cause of death in the country. The END Tuberculosis strategy stresses - the screening for symptoms alone may not suffice; additional screening tools such as a chest radiograph may facilitate referral for diagnosis of tuberculosis. The study aims to evaluate the diagnostic accuracy of artificial intelligence (AI) based Chest X-ray and compare it with the human reading (radiologist), using GeneXpert-MTB RIF Assay for tuberculosis case detection.
Methods: Tuberculosis-suspected patients with a history of cough were screened using chest X-rays at two study sites (Dhulikhel Hospital and Nobel Medical College). The reading of AI qXR software was compared with radiologists reading who were blinded of the results generated by the software.
Results: The sensitivity of the test by qXR-based AI reading was 100%, (95% CI: 40 – 100%) and specificity 80% (95% CI: 73 – 87%), whereas the sensitivity of the test by the radiologist was 100%, (95% CI: 40 – 100%); and specificity 62% (95% CI: 53 – 70%).
Conclusions: Higher sensitivity and specificity were observed for both qXR-based AI and Radiographer readings for the diagnosis of TB.
Keywords: Artificial Intelligence; chest X-ray; diagnostic accuracy; screening; tuberculosis.