AUTHOR=Yang Yang , Xia Lu , Liu Ping , Yang Fuping , Wu Yuqing , Pan Hongqiu , Hou Dailun , Liu Ning , Lu Shuihua 

TITLE=A prospective multicenter clinical research study validating the effectiveness and safety of a chest X-ray-based pulmonary tuberculosis screening software JF CXR-1 built on a convolutional neural network algorithm

JOURNAL=Frontiers in Medicine

VOLUME=Volume 10 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/medicine/articles/10.3389/fmed.2023.1195451

DOI=10.3389/fmed.2023.1195451

ISSN=2296-858X

ABSTRACT=Background: Chest radiography (chest X-ray, or CXR) has an important role in the early detection of active pulmonary tuberculosis (TB). While in areas with high-TB burden requiring screening urgently, radiologists are usually lacking. Computer-aided detection (CAD) software employed with artificial intelligence (AI) system may have the potential to solve this problem.Objective: We validated the effectiveness and safety of a pulmonary tuberculosis imaging screening software which is based on convolutional neural network algorithm.Methods: We conducted a prospective multicenter clinical research to validate the performance of a pulmonary tuberculosis imaging screening software (JF CXR-1).Volunteers with suspicious or without pulmonary tuberculosis aged no less than 15 years were recruited for CXR photography. The software reported a probability score of TB for each participant. The results were compared with those reported by radiologists. We measured sensitivity, specificity, consistency rate and the area under the receiver operating characteristic curves (AUC) for diagnosis of tuberculosis.Besides, adverse event (AE) and severe adverse event (SAE) were also evaluated.Results: The clinical research was conducted in six general infectious disease hospitals across mainland China. 1165 participants were enrolled and there were 1161 in the full analysis set (FAS). Males accounted for 60.0% (697/1161). Compared to the results from radiologists of the board, the software showed a sensitivity of 94.2% (95% CI: 92.0%-95.8%) and a specificity of 91.2% (95% CI: 88.5%-93.2%). The consistency rate was 92.7% (91.1%-94.1%) with a Kappa value of 0.854 (P=0.000).The AUC was 0.98. In the safety set (SS) which includes 1161 participants, 0.3% (3/1161) had AEs but were not related to the software, no severe AEs were observed.The software for tuberculosis screening based on convolutional neural network algorithm is effective and safe. It is a very potential candidate for solving the tuberculosis screening problem in areas lacking radiologists with high TB-burden.