Performance of ChatGPT-4o, Claude 3 Opus, and DeepSeek-R1 in BI-RADS Category 4 Classification and Malignancy Prediction From Mammography Reports: Retrospective Diagnostic Study
LLMs are feasible for distinguishing between benign and malignant lesions in BI-RADS category 4, with good stability and high sensitivity, but relatively insufficient specificity. They show potential in screening and may assist radiologists in reducing missed diagnoses.
Authors
- Yi Dai (ORCID: https://orcid.org/0000-0002-1919-1885)
- Shi Wei (ORCID: https://orcid.org/0009-0002-7690-1158)
- D. Xie (ORCID: https://orcid.org/0009-0007-4106-2477)
- Xingwei Dai (ORCID: https://orcid.org/0009-0007-7881-9648)
- Man Ke (ORCID: https://orcid.org/0009-0002-9832-4582)
- Mengting Mei (ORCID: https://orcid.org/0009-0004-7620-6862)
- Ronghua Yan (ORCID: https://orcid.org/0000-0003-3732-9638)
Institutions
- Southern University of Science and Technology (CN)
- Peking University Shenzhen Hospital (CN)
Publication Details
- Journal
- JMIR Medical Informatics
- Published
- 2025-11-30
- DOI
- https://doi.org/10.2196/80182
- Primary Topic
- Radiomics and Machine Learning in Medical Imaging
- Type
- article
- Field-Weighted Citation Impact
- 0.00