The Lancet: Digital Health (Sep 2023)

Pan-mediastinal neoplasm diagnosis via nationwide federated learning: a multicentre cohort study

  • Ruijie Tang, PhD,
  • Hengrui Liang, PhD,
  • Yuchen Guo, PhD,
  • Zhigang Li, PhD,
  • Zhichao Liu, MD,
  • Xu Lin, PhD,
  • Zeping Yan, BS,
  • Jun Liu, PhD,
  • Xin Xu, PhD,
  • Wenlong Shao, PhD,
  • Shuben Li, PhD,
  • Wenhua Liang, ProfPhD,
  • Wei Wang, PhD,
  • Fei Cui, PhD,
  • Huanghe He, PhD,
  • Chao Yang, PhD,
  • Long Jiang, PhD,
  • Haixuan Wang, MD,
  • Huai Chen, MD,
  • Chenguang Guo, MD,
  • Haipeng Zhang, MD,
  • Zebin Gao, PhD,
  • Yuwei He, PhD,
  • Xiangru Chen, ME,
  • Lei Zhao, PhD,
  • Hong Yu, MD,
  • Jian Hu, ProfPhD,
  • Jiangang Zhao, MD,
  • Bin Li, PhD,
  • Ci Yin, MD,
  • Wenjie Mao, MD,
  • Wanli Lin, MD,
  • Yujie Xie, MD,
  • Jixian Liu, MD,
  • Xiaoqiang Li, PhD,
  • Dingwang Wu, MD,
  • Qinghua Hou, MD,
  • Yongbing Chen, PhD,
  • Donglai Chen, PhD,
  • Yuhang Xue, MD,
  • Yi Liang, MD,
  • Wenfang Tang, MD,
  • Qi Wang, ProfPhD,
  • Encheng Li, MD,
  • Hongxu Liu, MD,
  • Guan Wang, MD,
  • Pingwen Yu, MD,
  • Chun Chen, ProfPhD,
  • Bin Zheng, PhD,
  • Hao Chen, MD,
  • Zhe Zhang, MD,
  • Lunqing Wang, MD,
  • Ailin Wang, MD,
  • Zongqi Li, MD,
  • Junke Fu, MD,
  • Guangjian Zhang, MD,
  • Jia Zhang, MD,
  • Bohao Liu, MD,
  • Jian Zhao, MD,
  • Boyun Deng, MD,
  • Yongtao Han, PhD,
  • Xuefeng Leng, PhD,
  • Zhiyu Li, MD,
  • Man Zhang, MD,
  • Changling Liu, MD,
  • Tianhu Wang, MD,
  • Zhilin Luo, MD,
  • Chenglin Yang, MD,
  • Xiaotong Guo, MD,
  • Kai Ma, MD,
  • Lixu Wang, MD,
  • Wenjun Jiang, MD,
  • Xu Han, MD,
  • Qing Wang, MD,
  • Kun Qiao, MD,
  • Zhaohua Xia, MD,
  • Shuo Zheng, MD,
  • Chenyang Xu, MD,
  • Jidong Peng, MD,
  • Shilong Wu, MD,
  • Zhifeng Zhang, MD,
  • Haoda Huang, MD,
  • Dazhi Pang, MD,
  • Qiao Liu, BSN,
  • Jinglong Li, MD,
  • Xueru Ding, MD,
  • Xiang Liu, MD,
  • Liucheng Zhong, MD,
  • Yutong Lu, MD,
  • Feng Xu, PhD,
  • Qionghai Dai, ProfPhD,
  • Jianxing He, ProfPhD

Journal volume & issue
Vol. 5, no. 9
pp. e560 – e570

Abstract

Read online

Summary: Background: Mediastinal neoplasms are typical thoracic diseases with increasing incidence in the general global population and can lead to poor prognosis. In clinical practice, the mediastinum's complex anatomic structures and intertype confusion among different mediastinal neoplasm pathologies severely hinder accurate diagnosis. To solve these difficulties, we organised a multicentre national collaboration on the basis of privacy-secured federated learning and developed CAIMEN, an efficient chest CT-based artificial intelligence (AI) mediastinal neoplasm diagnosis system. Methods: In this multicentre cohort study, 7825 mediastinal neoplasm cases and 796 normal controls were collected from 24 centres in China to develop CAIMEN. We further enhanced CAIMEN with several novel algorithms in a multiview, knowledge-transferred, multilevel decision-making pattern. CAIMEN was tested by internal (929 cases at 15 centres), external (1216 cases at five centres and a real-world cohort of 11 162 cases), and human–AI (60 positive cases from four centres and radiologists from 15 institutions) test sets to evaluate its detection, segmentation, and classification performance. Findings: In the external test experiments, the area under the receiver operating characteristic curve for detecting mediastinal neoplasms of CAIMEN was 0·973 (95% CI 0·969–0·977). In the real-world cohort, CAIMEN detected 13 false-negative cases confirmed by radiologists. The dice score for segmenting mediastinal neoplasms of CAIMEN was 0·765 (0·738–0·792). The mediastinal neoplasm classification top-1 and top-3 accuracy of CAIMEN were 0·523 (0·497–0·554) and 0·799 (0·778–0·822), respectively. In the human–AI test experiments, CAIMEN outperformed clinicians with top-1 and top-3 accuracy of 0·500 (0·383–0·633) and 0·800 (0·700–0·900), respectively. Meanwhile, with assistance from the computer aided diagnosis software based on CAIMEN, the 46 clinicians improved their average top-1 accuracy by 19·1% (0·345–0·411) and top-3 accuracy by 13·0% (0·545–0·616). Interpretation: For mediastinal neoplasms, CAIMEN can produce high diagnostic accuracy and assist the diagnosis of human experts, showing its potential for clinical practice. Funding: National Key R&D Program of China, National Natural Science Foundation of China, and Beijing Natural Science Foundation.