Comparing the Output of an Artificial Intelligence Algorithm in Detecting Radiological Signs of Pulmonary Tuberculosis in Digital Chest X-Rays and Their Smartphone-Captured Photos of X-Ray Films: Retrospective Study

This study aimed to compare the sensitivity and specificity of AI in detecting radiological signs of tuberculosis (TB) using digital chest X-ray (CXR) DICOM files versus smartphone-captured photos of digital CXR films. A total of 1,278 CXR pairs were analyzed, with AI results compared against a radiological ground truth established by a panel of three radiologists. AI demonstrated a sensitivity of 92.22% (95% CI: 89.94–94.12) for digital CXRs and 90.75% (95% CI: 88.32–92.82) for analog CXRs (P = 0.09). The specificity was 82.08% (95% CI: 78.76–85.07) for digital CXRs and 79.23% (95% CI: 75.75–82.42) for analog CXRs (P = 0.06), indicating no statistically significant difference. The findings suggest that AI can effectively analyze both digital and analog CXR images for TB detection, making it a viable tool in settings where only printed CXR films are available.