An expert-level generalist AI for abdominal CT diagnosis
Artificial intelligence (AI) in radiology aspires to deliver expert-level diagnosis across diverse clinical tasks, yet existing supervised strategies remain limited in scope. We developed RADAR, a generalist vision-language model trained on more than 400,000 contrast-enhanced abdominal computed tomography (CT) examinations and 15 million anatomy-wise image-text pairs, learning directly from clinical reports without manual annotation. Throughout internal and external evaluations across multiple centers and varied clinical scenarios, RADAR achieved high diagnostic performance and robust generalization for 18 anatomical structures and 146 imaging findings. In a reader study, RADAR assistance increased the diagnostic sensitivity of 26 radiologists by ~10%. RADAR offers a scalable, versatile, and interpretable solution for abdominal CT, demonstrating that generalist AI can match human experts in general and complicated radiology tasks.
Authors
- He Ding (ORCID: https://orcid.org/0000-0002-2056-4618)
- Shenghong Ju (ORCID: https://orcid.org/0000-0001-5041-7865)
- Yutong Xie (ORCID: https://orcid.org/0000-0002-6644-1250)
- Hongkan Wang
- Jian Ding (ORCID: https://orcid.org/0000-0001-8419-273X)
- Jianwen Ning
- Shaoteng Zhang
- Zhilin Zheng (ORCID: https://orcid.org/0000-0002-2439-9162)
- Xianghua Ye (ORCID: https://orcid.org/0000-0002-4075-4777)
- Ling Zhang (ORCID: https://orcid.org/0000-0001-8371-5252)
- Tony C. W. Mok (ORCID: https://orcid.org/0000-0002-4779-9337)
- Wanxing Chang
- Yanjie Zhou (ORCID: https://orcid.org/0000-0003-2222-9140)
- Zhongyi Shui (ORCID: https://orcid.org/0009-0009-8052-7972)
- Yingda Xia (ORCID: https://orcid.org/0000-0002-7478-4392)
- Tao Ma (ORCID: https://orcid.org/0000-0002-0579-8045)
- Chaohui Yu (ORCID: https://orcid.org/0000-0003-4842-3646)
- Xiaoguang Wang (ORCID: https://orcid.org/0000-0002-4079-9596)
- Xi Li (ORCID: https://orcid.org/0000-0003-3023-1662)
- Jian Liu (ORCID: https://orcid.org/0000-0001-7663-5844)
- Zilin Lu (ORCID: https://orcid.org/0000-0003-2437-283X)
- Yong Xia (ORCID: https://orcid.org/0000-0001-9273-2847)
- Xing Xue (ORCID: https://orcid.org/0000-0002-1154-1393)
- Qi Zhang (ORCID: https://orcid.org/0000-0002-6096-0690)
- Wenbo Xiao (ORCID: https://orcid.org/0000-0001-7124-1096)
- Weiwei Cao (ORCID: https://orcid.org/0000-0002-8991-9915)
- Jie Peng (ORCID: https://orcid.org/0000-0002-4005-085X)
- Sinuo Wang
- Tingbo Liang (ORCID: https://orcid.org/0000-0003-0143-3353)
- Cao Chen (ORCID: https://orcid.org/0000-0002-2183-9855)
- Yiping Liu
- Jianfeng Zhang
- Dongjie Chen
- Wei Zhang
- Jianpeng Zhang (ORCID: https://orcid.org/0009-0001-5077-0500)
- Qi Wu
- Zhaomin Ni (ORCID: https://orcid.org/0009-0002-1640-2955)
- Huazhen Ye
- Yuming Gao
- Zhi Li
Institutions
- University of Nottingham Ningbo China (CN)
- Zhejiang University of Science and Technology (CN)
- Xinjiang Production and Construction Corps (CN)
- Australian Institute of Business (AU)
- Alibaba Group (Cayman Islands) (KY)
- Jiaxing University (CN)
- First People's Hospital of Yuhang District (CN)
- Ningbo No. 2 Hospital (CN)
- Zhongda Hospital Southeast University (CN)
- Mohamed bin Zayed University of Artificial Intelligence (AE)
- Zhejiang Lab (CN)
- Heilongjiang University of Technology (CN)
- First Hospital of Jiaxing (CN)
- Third People's Hospital of Huzhou (CN)
- Shaoxing People's Hospital (CN)
- Jingning County People's Hospital (CN)
- First Affiliated Hospital Zhejiang University (CN)
- Alibaba Group (China) (CN)
- The University of Adelaide (AU)
- Zhejiang University (CN)
Publication Details
- Journal
- Science
- Published
- 2026-09-17
- DOI
- https://doi.org/10.1126/science.aec6129
- Primary Topic
- Artificial Intelligence in Healthcare and Education
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- National Natural Science Foundation of China