Fine-Tuned Large Language Models for Automated Radiology Impression Generation: A Multicenter Evaluation
< .001). Conclusion MIRA, a fine-tuned LLM using a prompt-based strategy, generated clinically aligned radiology impressions in multicenter settings, improving accuracy, efficiency, and reporting consistency. © The Authors 2026. Published by the Radiological Society of North America under a CC BY 4.0 license.
Authors
- Lijun Duan
- Han Xue
- Lei Zhang (ORCID: https://orcid.org/0000-0002-5567-7647)
- Jing Wang (ORCID: https://orcid.org/0000-0001-6675-361X)
- Jiaqi Gong
- Miao Zheng (ORCID: https://orcid.org/0000-0003-3269-584X)
- Lin Mu (ORCID: https://orcid.org/0000-0002-7126-3938)
- Lu Liu (ORCID: https://orcid.org/0000-0001-7712-1765)
- Yunhui Yang
- Qi Dai
- Munire Aihemaiti (ORCID: https://orcid.org/0009-0005-2979-3395)
- Kai Zhu
- Ying Mu
- Mingyang Li (ORCID: https://orcid.org/0000-0002-9642-3133)
Institutions
- Fujian Medical University (CN)
- Harbin Medical University (CN)
- Jilin University (CN)
- Ningbo No. 2 Hospital (CN)
- Gaochun People's Hospital (CN)
- The Central Hospital of Enshi Tujia and Miao Autonomous Prefecture (CN)
- First Affiliated Hospital of Fujian Medical University (CN)
- First Hospital of Jilin University (CN)
- Jiangsu Province Hospital (CN)
- First Affiliated Hospital of Harbin Medical University (CN)
- Ningbo No.6 Hospital (CN)
- China Medical University (CN)
Publication Details
- Journal
- Radiology Artificial Intelligence
- Published
- 2026-04-15
- DOI
- https://doi.org/10.1148/ryai.250714
- Primary Topic
- Artificial Intelligence in Healthcare and Education
- Type
- article
- Field-Weighted Citation Impact
- 0.00