Diagnostic and interpretive gains from reasoning over conclusions with a large reasoning model in radiology
Ruixin Wang, Jinghang Wang, Yisong Wang, Chao Zheng, Sihong Huang, Xiaohui Liu, Guoping Tan, Wei Zhao, Zhiyuan Wang, Shaoliang Peng, Jun Liu · 2025
Read Diagnostic and interpretive gains from reasoning over conclusions with a large reasoning model in radiology by Ruixin Wang, Jinghang Wang, Yisong Wang, Chao Zheng, Sihong Huang, Xiaohui Liu, Guoping Tan, Wei Zhao, Zhiyuan Wang, Shaoliang Peng, Jun Liu (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-12858915.txt.
Description
Ruixin Wang and co-authors test a large reasoning model on radiology work in a 2025 study. Radiologists can miss subtle second findings that matter for tumor staging and care. Reasoning models, which spell out their steps before a final answer, might catch more and show their work. Yet no one had checked that claim in a systematic way.
The team studied nine hundred cancer cases from three Chinese hospitals. They compared the model's step-by-step reading with its answer-only form and with two models that do not reason aloud. Three senior radiologists graded errors and quality traits. The reasoning form made the fewest missed or wrong calls and scored best on completeness and clarity, though it was less concise. A side study with six radiologists tested workflow effects, and an English data set tested language transfer.
Catalog record
| CREATOR | Ruixin Wang, Jinghang Wang, Yisong Wang, Chao Zheng, Sihong Huang, Xiaohui Liu, Guoping Tan, Wei Zhao, Zhiyuan Wang, Shaoliang Peng, Jun Liu |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 56 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |