The accuracy of mobile teleradiology in the evaluation of chest X-rays

We assessed the diagnostic accuracy of digital photographs of plain film chest X-rays (CXRs) obtained using a mobile phone. The study was a randomized, non-inferiority trial, in which physical plain film CXRs viewed on a light box were compared with digital photographs of plain film CXRs. CXRs were selected from a database of radiology studies to show common pathologies found in Botswana associated with pneumonia, lung carcinoma, tuberculosis, pneumothorax and interstitial disease, as well as normal findings. The pre-selected diagnoses were subsequently verified by a second radiologist. Seven radiologists were randomized to review 75 plain film CXRs on light boxes before viewing 75 digital photographs, or vice versa. Their responses were considered correct if they matched the pre-defined diagnosis. For both modalities, the correct diagnosis was provided in 79% of cases; for plain film CXRs, the correct diagnosis was provided in 82% of cases and for digital photographs the correct diagnosis was provided in 76% of cases. The difference in diagnostic accuracy was -5.7% (95% CI: -10.8% to -0.5%), which confirmed non-inferiority (P<0.001) for the primary outcome of diagnostic accuracy. A subgroup analysis demonstrated non-inferiority for lung carcinoma and pneumonia images, although non-inferiority was not achieved for pneumothorax, tuberculosis, interstitial disease or normal images. The study demonstrates that digital photographs of CXRs obtained via a mobile phone equipped with a digital camera are non-inferior to plain film CXRs.