Chest X-ray Abnormality Detection by Using Artificial Intelligence: A Single-Site Retrospective Study of Deep Learning Model Performance

Chest X-ray (CXR) is one of the most common radiological examinations for both nonemergent and emergent clinical indications, but human error or lack prioritization patients can hinder timely interpretation. Deep learning (DL) algorithms have proven to be useful in assessment various abnormalities including tuberculosis, lung parenchymal lesions, pneumothorax. The deep learning–based automatic detection algorithm (DLAD) was developed detect visual patterns on CXR 12 preselected findings. To evaluate proposed system, we designed a single-site retrospective study comparing DL with performance five differently experienced radiologists. On assessed dataset (n = 127) collected from municipal hospital Czech Republic, DLAD achieved sensitivity (Se) 0.925 specificity (Sp) 0.644, compared bootstrapped radiologists’ Se 0.661 Sp 0.803, respectively, statistically significant difference. negative likelihood ratio (NLR) software (0.12 (0.04–0.32)) significantly lower than (0.42 (0.4–0.43), p < 0.0001). No critical findings were missed by software.