Author:
Chen Chiu-Fan,Hsu Chun-Hsiang,Jiang You-Cheng,Lin Wen-Ren,Hong Wei-Cheng,Chen I.-Yuan,Lin Min-Hsi,Chu Kuo-An,Lee Chao-Hsien,Lee David Lin,Chen Po-Fan
Abstract
AbstractIn tuberculosis (TB), chest radiography (CXR) patterns are highly variable, mimicking pneumonia and many other diseases. This study aims to evaluate the efficacy of Google teachable machine, a deep neural network-based image classification tool, to develop algorithm for predicting TB probability of CXRs. The training dataset included 348 TB CXRs and 3806 normal CXRs for training TB detection. We also collected 1150 abnormal CXRs and 627 normal CXRs for training abnormality detection. For external validation, we collected 250 CXRs from our hospital. We also compared the accuracy of the algorithm to five pulmonologists and radiological reports. In external validation, the AI algorithm showed areas under the curve (AUC) of 0.951 and 0.975 in validation dataset 1 and 2. The accuracy of the pulmonologists on validation dataset 2 showed AUC range of 0.936–0.995. When abnormal CXRs other than TB were added, AUC decreased in both human readers (0.843–0.888) and AI algorithm (0.828). When combine human readers with AI algorithm, the AUC further increased to 0.862–0.885. The TB CXR AI algorithm developed by using Google teachable machine in this study is effective, with the accuracy close to experienced clinical physicians, and may be helpful for detecting tuberculosis by CXR.
Publisher
Springer Science and Business Media LLC
Reference26 articles.
1. World Health Organization. Global Tuberculosis Report 2021. https://www.who.int/publications/i/item/9789240037021 (World Health Organization, 2021).
2. Perez-Guzman, C., Torres-Cruz, A., Villarreal-Velarde, H. & Vargas, M. H. Progressive age-related changes in pulmonary tuberculosis images and the effect of diabetes. Am. J. Respir. Crit. Care Med. 162(5), 1738–1740. https://doi.org/10.1164/ajrccm.162.5.2001040 (2000).
3. Mathur, M., Badhan, R. K., Kumari, S., Kaur, N. & Gupta, S. Radiological manifestations of pulmonary tuberculosis—A comparative study between immunocompromised and immunocompetent patients. J. Clin. Diagn. Res. 11(9), TC06–TC09. https://doi.org/10.7860/JCDR/2017/28183.10535 (2017).
4. Rajpurkar, P. et al. Deep learning for chest radiograph diagnosis: A retrospective comparison of the CheXNeXt algorithm to practicing radiologists. PLoS Med. 15(11), e1002686. https://doi.org/10.1371/journal.pmed.1002686 (2018).
5. Cohen, J. P., Bertin, P., & Frappier, V. Chester: A Web Delivered Locally Computed Chest X-Ray Disease Prediction System. arXiv: https://arxiv.org/abs/1901.11210 (2020).