Xin Henan, Wang Wei, Wang Xiaomeng, Huang Juanjuan, Di Yuanzhi, Du Jiang, Cao Xuefang, Feng Boxuan, Shen Lingyu, He Yijun, Guo Tonglei, Li Zihan, Liang Jianguo, Wang Zhen, Zhu Ping, Gao Lei
NHC Key Laboratory of Systems Biology of Pathogens, National Institute of Pathogen Biology, and Center for Tuberculosis Research, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, People's Republic of China.
Key Laboratory of Pathogen Infection Prevention and Control (Ministry of Education), National Institute of Pathogen Biology, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing, People's Republic of China.
Emerg Microbes Infect. 2025 Dec;14(1):2470998. doi: 10.1080/22221751.2025.2470998. Epub 2025 Apr 28.
From 2020 to 2022, a pulmonary tuberculosis (PTB) active case finding project based on chest X-ray (CXR) examination was conducted targeting individuals aged ≥65 years old in Jiangshan County, Quzhou City. The current study used computer-aided detection (CAD) software (JF CXR-1 v2) to retrospectively analyze the CXR images and to estimate its potential capacity for identifying PTB cases. The information of notified microbiologically confirmed PTB among the participants were exported from the Tuberculosis Information Management System. A total of 49,919 subjects participated in the 2020 examinations. Of these, 40,741 and 39,185 completed the follow-up surveys in 2021 and 2022, respectively. The pooled prevalence of suspected PTB reported by radiologists was 1.21% (1579/129,776), compared with 12.43% (16,129/129,776) reported by CAD. Of 101 bacteriologically confirmed PTB cases notified over three years, radiologists and CAD reported 45.54% (46/101) and 83.16% (84/101) as suspected cases, respectively. Among subjects with abnormal CAD (CAD score>0.35), the majority of the notified confirmed PTB patients (63/84) had their CAD scores >75% quantiles (as>0.75). With 3 years' results, their CAD scores exhibited dynamic changes along with disease progression or treatment with median scores peaking in the year of diagnosis. This intriguing finding suggests that CAD for CXR reading assisted radiologists in PTB screening by reducing workload and improving case finding. The CAD primary score may have the potential to identify high-risk individuals and early PTB patients, adding a new dimension to our understanding of disease progression.
2020年至2022年，在衢州市江山县针对65岁及以上人群开展了一项基于胸部X光（CXR）检查的肺结核（PTB）主动病例发现项目。本研究使用计算机辅助检测（CAD）软件（JF CXR - 1 v2）对CXR图像进行回顾性分析，并评估其识别PTB病例的潜在能力。参与者中经微生物学确诊的PTB通报信息从结核病信息管理系统导出。共有49,919名受试者参加了2020年的检查。其中，分别有40,741名和39,185名在2021年和2022年完成了随访调查。放射科医生报告的疑似PTB合并患病率为1.21%（1579/129,776），而CAD报告的患病率为12.43%（16,129/129,776）。在三年中通报的101例经细菌学确诊的PTB病例中，放射科医生和CAD分别将45.54%（46/101）和83.16%（84/101）报告为疑似病例。在CAD异常（CAD评分>0.35）的受试者中，大多数通报确诊的PTB患者（63/84）的CAD评分高于第75百分位数（>0.75）。根据三年的结果，他们的CAD评分随着疾病进展或治疗呈现动态变化，中位数评分在诊断年份达到峰值。这一有趣的发现表明，用于CXR阅片的CAD通过减轻工作量和改善病例发现，辅助放射科医生进行PTB筛查。CAD初始评分可能有潜力识别高危个体和早期PTB患者，为我们对疾病进展的理解增添了新的维度。
