T-SPOT with CT image analysis based on deep learning for early differential diagnosis of nontuberculous mycobacteria pulmonary disease and pulmonary tuberculosis

• DOI: 10.1016/j.ijid.2022.09.031
• Publication Date: 2022-09-28T05:13:39Z
• Authors: Chiqing Ying; Xukun Li; Shuangzhi Lv; Peng Du; Yunzhi Chen; Hongxin Fu; Weibo Du; Kaijin Xu; Ying Zhang; Wei Wu
• Keywords: Nontuberculous Mycobacteria

This study aimed to establish a diagnostic algorithm combining T-SPOT with computed tomography image analysis based on deep learning (DL) for early differential diagnosis of nontuberculous mycobacteria pulmonary disease (NTM-PD) and tuberculosis (PTB).A total 1049 cases were enrolled, including 467 NTM-PD 582 PTB cases. A 320 (160 160 PTB) randomized as the testing set analyzed using combined DL model. The first divided into T-SPOT-positive -negative groups, model was then used separate four subgroups further.The precision found be 91.7% subgroup T-SPOT-negative classified NTM-PD, 89.8% PTB, which covered 66.9% cases, compared accuracy rate 80.3% alone. In other two remaining where prediction inconsistent model, 73.0% 52.2%, separately.Our shows that new system can greatly improve classification when methods are consistent.