A deep-learning-based framework for identifying and localizing multiple abnormalities and assessing cardiomegaly in chest X-ray

作者全名："Fan, Weijie; Yang, Yi; Qi, Jing; Zhang, Qichuan; Liao, Cuiwei; Wen, Li; Wang, Shuang; Wang, Guangxian; Xia, Yu; Wu, Qihua; Fan, Xiaotao; Chen, Xingcai; He, Mi; Xiao, JingJing; Yang, Liu; Liu, Yun; Chen, Jia; Wang, Bing; Zhang, Lei; Yang, Liuqing; Gan, Hui; Zhang, Shushu; Liu, Guofang; Ge, Xiaodong; Cai, Yuanqing; Zhao, Gang; Zhang, Xi; Xie, Mingxun; Xu, Huilin; Zhang, Yi; Chen, Jiao; Li, Jun; Han, Shuang; Mu, Ke; Xiao, Shilin; Xiong, Tingwei; Nian, Yongjian; Zhang, Dong"

作者地址："[Fan, Weijie; Zhang, Qichuan; Liao, Cuiwei; Wen, Li; Wang, Shuang; Yang, Liu; Liu, Yun; Chen, Jia; Wang, Bing; Zhang, Lei; Yang, Liuqing; Gan, Hui; Zhang, Shushu; Liu, Guofang; Ge, Xiaodong; Cai, Yuanqing; Zhao, Gang; Zhang, Xi; Xie, Mingxun; Xu, Huilin; Zhang, Yi; Chen, Jiao; Li, Jun; Han, Shuang; Mu, Ke; Xiao, Shilin; Xiong, Tingwei; Zhang, Dong] Army Med Univ, Affiliated Hosp 2, Dept Radiol, Chongqing 400037, Peoples R China; [Yang, Yi; Qi, Jing; Chen, Xingcai; He, Mi; Nian, Yongjian] Army Med Univ, Sch Biomed Engn & Imaging Med, Dept Digital Med, Chongqing 400038, Peoples R China; [Wang, Guangxian] Chongqing Med Univ, Peoples Hosp Banan, Dept Radiol, Chongqing 401320, Peoples R China; [Xia, Yu] Xishui Hosp Tradit Chinese Med, Dept Radiol, Zunyi 564600, Zunyi Of Guizho, Peoples R China; [Wu, Qihua] Peoples Hosp Nanchuan, Dept Radiol, Chongqing 408400, Peoples R China; [Fan, Xiaotao] Fengdu Peoples Hosp, Dept Radiol, Chongqing 408200, Peoples R China; [Xiao, JingJing] Army Med Univ, Affiliated Hosp 2, Dept Med Engn, Chongqing 400037, Peoples R China"

通信作者："Zhang, D (通讯作者)，Army Med Univ, Affiliated Hosp 2, Dept Radiol, Chongqing 400037, Peoples R China.; Nian, YJ (通讯作者)，Army Med Univ, Sch Biomed Engn & Imaging Med, Dept Digital Med, Chongqing 400038, Peoples R China."

来源：NATURE COMMUNICATIONS

ESI学科分类：　

WOS号：WOS:001162717700006

JCR分区：Q1

影响因子：14.7

年份：2024

卷号：15

期号：1

开始页：　

结束页：　

文献类型：Article

关键词：　

摘要："Accurate identification and localization of multiple abnormalities are crucial steps in the interpretation of chest X-rays (CXRs); however, the lack of a large CXR dataset with bounding boxes severely constrains accurate localization research based on deep learning. We created a large CXR dataset named CXR-AL14, containing 165,988 CXRs and 253,844 bounding boxes. On the basis of this dataset, a deep-learning-based framework was developed to identify and localize 14 common abnormalities and calculate the cardiothoracic ratio (CTR) simultaneously. The mean average precision values obtained by the model for 14 abnormalities reached 0.572-0.631 with an intersection-over-union threshold of 0.5, and the intraclass correlation coefficient of the CTR algorithm exceeded 0.95 on the held-out, multicentre and prospective test datasets. This framework shows an excellent performance, good generalization ability and strong clinical applicability, which is superior to senior radiologists and suitable for routine clinical settings. Accurate localization of abnormalities is crucial in the interpretation of chest X-rays. Here the authors present a deep learning framework for simultaneous localization of 14 thoracic abnormalities and calculation of cardiothoracic ratio, based on large X-ray dataset with bounding boxes created via a human-in-the-loop approach."

基金机构：Clinical major innovative characteristic technology project of Second Affiliated Hospital of Army Medical University [2018JSLC0016]; Talent project of Chongqing [CQYC202103075]

基金资助正文：This work was supported by the clinical major innovative characteristic technology project of Second Affiliated Hospital of Army Medical University (2018JSLC0016 for D.Z.) and by the Talent project of Chongqing (CQYC202103075 for D.Z.). We thank the information department in Second Affiliated Hospital of Army Medical University for providing extensive CXRs data and all those who had made contribution to the publication of the work.