CT-based artificial intelligence system complementing deep learning model and radiologist for liver fibrosis staging
Shuang Zheng,
Wenao Ma,
Lin Mu,
Kan He,
Jianfeng Cao,
Tiffany Y. So,
Lei Zhang,
Mingyang Li,
Yanan Zhai,
Feng Liu,
Shunlin Guo,
Longlin Yin,
Liming Zhao,
Lei Wang,
Heather H.C. Lee,
Wei Jiang,
Junqi Niu,
Pujun Gao,
Qi Dou,
Huimao Zhang
Affiliations
Shuang Zheng
Department of Radiology, The First Hospital of Jilin University, Changchun, China
Wenao Ma
Department of Computer Science and Engineering, The Chinese University of Hong Kong, Hong Kong, China
Lin Mu
Department of Radiology, The First Hospital of Jilin University, Changchun, China
Kan He
Department of Radiology, The First Hospital of Jilin University, Changchun, China
Jianfeng Cao
Department of Computer Science and Engineering, The Chinese University of Hong Kong, Hong Kong, China
Tiffany Y. So
Department of Imaging and Interventional Radiology, Prince of Wales Hospital, Hong Kong, China
Lei Zhang
Department of Radiology, The First Hospital of Jilin University, Changchun, China
Mingyang Li
Department of Radiology, The First Hospital of Jilin University, Changchun, China
Yanan Zhai
Department of Radiology, The First Hospital of Lanzhou University, The First Clinical Medical College of Lanzhou University, Intelligent Imaging Medical Engineering Research Center of Gansu Province, Radiological Clinical Medicine Research Center of Gansu Province, Lanzhou, China
Feng Liu
The First Clinical Medical College of Lanzhou University, Lanzhou, China
Shunlin Guo
Department of Radiology, The First Hospital of Lanzhou University, The First Clinical Medical College of Lanzhou University, Intelligent Imaging Medical Engineering Research Center of Gansu Province, Radiological Clinical Medicine Research Center of Gansu Province, Lanzhou, China
Longlin Yin
Department of Radiology, Sichuan Academy of Medical Sciences & Sichuan Provincial People’s Hospital, Chengdu, China
Liming Zhao
Department of Radiology, Sichuan Academy of Medical Sciences & Sichuan Provincial People’s Hospital, Chengdu, China
Lei Wang
Department of Imaging and Interventional Radiology, Prince of Wales Hospital, Hong Kong, China
Heather H.C. Lee
Department of Diagnostic Radiology, Princess Margaret Hospital, Hong Kong, China
Wei Jiang
National Health Commission Capacity Building and Continuing Education Center, Department of Big Data, Beijing, China
Junqi Niu
Department of Hepatology, The First Hospital of Jilin University, Changchun, China
Pujun Gao
Department of Hepatology, The First Hospital of Jilin University, Changchun, China
Qi Dou
Department of Computer Science and Engineering, The Chinese University of Hong Kong, Hong Kong, China; Corresponding author
Huimao Zhang
Department of Radiology, The First Hospital of Jilin University, Changchun, China; Corresponding author
Summary: Noninvasive methods for liver fibrosis staging are urgently needed due to its significance in predicting significant morbidity and mortality. In this study, we developed an automated DL-based segmentation and classification model (Model-C). Test-time adaptation was used to address data distribution shifts. We then established a deep learning-radiologist complementarity decision system (DRCDS) via a decision model determining whether to adopt Model-C’s diagnosis or defer to radiologists. Model-C (AUCs of 0.89–0.92) outperformed models based on liver (AUCs: 0.84–0.90) or spleen (AUCs: 0.69–0.70). With test-time adaptation, the Obuchowski index values of Model-C in three external sets improved from 0.81, 0.73, and 0.73 to 0.85, 0.85, and 0.81. DRCDS performed slightly better than Model-C or senior radiologists, with 73.7%–92.0% of cases adopting Model-C’s diagnosis. In conclusion, DRCDS could diagnose liver fibrosis with high accuracy. Additionally, we provided solutions to model generalization and human-machine complementarity issues in multi-classification problems.
