Mobile phone-based computer-aided chest X-ray detects TB as well as trained radiologists

Background: Chest radiography (CXR) is the most common imaging modality used to diagnose pulmonary tuberculosis (PTB). Accurate computer-aided detection (CAD) CXR software may aid physicians in low-resource, high tuberculosis (TB) endemic settings where radiologists are scarce. Methods: A retrospective pilot study comparing the assessment of humans and CAD was conducted on analogue CXRs from Guinea-Bissau and Ethiopia with the objective of comparing the interpretation of CXRs regarding PTB by CAD (qXR, Qure.ai) with that of two experienced Ethiopian radiologists (A and B). To improve the applicability of this method in low-resource settings, an analysis was performed on images of CXRs taken by mobile phones. Two reference standards were applied: final TB diagnosis by clinical or laboratory findings (i.e., GeneXpert (GX)-confirmed TB). Results: We included 498 CXRs. Radiologist A identified 50, radiologist B identified 99, and the software identified 81 as indicative of TB. The overall area under the curve (AUC) for the receiver operating characteristic (ROC) curve of the software was 0.84 for GX-confirmed cases. At the prechosen cut-off value of 0.5, the sensitivity of CAD CXR was 76.5%, and the specificity was 85.9%. Radiologist A’s assessments were 64.7% sensitive and 91.9% specific, while radiologist B´s assessments were 76.5% sensitive and 82.3% specific for GX-confirmed cases. The agreement regarding TB-related findings between the radiologists combined (k=0.45) and each radiologist and the software (k=0.56) was moderate. Conclusion: Our study revealed that CAD CXR performs comparably to experienced radiologists in diagnosing TB even when it is applied to analogue CXRs photographed by mobile phones. Trial registration: PACTR201611001838365