EUVIMEDEUVIMED entdecken
Crossref

FiR-Rad: Fine-Grained Reinforcement With Structured Reasoning for Chest X-Ray Report Generation

Xin Mei, Libin Yang, Dehong Gao, Xiaoyan Cai, Junwei Han, Tianming Liu

IEEE Transactions on Medical Imaging · 2026-06 · 45 (6) · 2922-2935

DOI: 10.1109/tmi.2026.3665561 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Medical Imaging, 2026-06. Verfasst von Xin Mei und 5 weiteren Autor:innen. Hauptthemen laut MEDLINE: Radiography, Thoracic, Radiographic Image Interpretation, Computer-Assisted, Radiology Information Systems. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 57 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Key Research and Development Plan of Shaanxi Province, National Key Research and Development Project.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Automated chest X-ray report generation requires not only clinical accuracy but also transparent and interpretable diagnostic reasoning. In this work, we propose FiR-Rad, a two-stage framework that combines explicit structured reasoning with targeted fine-grained optimization. In the first stage, a supervised chain-of-thought approach guides the model to sequentially analyze and describe a comprehensive range of clinically significant thoracic abnormalities, ensuring clinically meaningful coverage. In the second stage, we introduce a segment-level reinforcement learning strategy based on Group Relative Policy Optimization (GRPO), which assigns precise rewards to each disease-specific reasoning step by evaluating the accuracy of corresponding findings in the synthesized report. This design provides direct feedback for intermediate reasoning and encourages consistency between detailed abnormality analysis and final diagnostic conclusions. Experimental results on the MIMIC-CXR and IU-Xray datasets demonstrate that our framework achieves state-of-the-art performance across clinical and linguistic metrics, with strong zero-shot generalization on IU-Xray. The proposed method significantly enhances interpretability and clinical accuracy, effectively addressing key limitations in automated radiology report generation.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Radiography, ThoracicRadiographic Image Interpretation, Computer-AssistedRadiology Information SystemsHumansReinforcement Machine LearningAlgorithms

Förderung

  • National Natural Science Foundation of China
  • Key Research and Development Plan of Shaanxi Province
  • National Key Research and Development Project
  • Natural Science Basic Research Program of Shaanxi
  • CCF-NSFOCUS ‘Kunpeng’ Research Fund
  • Huawei Research Project

Zitieren

APA

Mei, X., Yang, L., Gao, D., Cai, X., Han, J., & Liu, T. (2026). FiR-Rad: Fine-Grained Reinforcement With Structured Reasoning for Chest X-Ray Report Generation. IEEE Transactions on Medical Imaging, 45(6), 2922-2935. https://doi.org/10.1109/tmi.2026.3665561

Vancouver

Mei X, Yang L, Gao D, Cai X, Han J, Liu T. FiR-Rad: Fine-Grained Reinforcement With Structured Reasoning for Chest X-Ray Report Generation. IEEE Transactions on Medical Imaging. 2026;45(6):2922-2935. doi:10.1109/tmi.2026.3665561

BibTeX
@article{Mei2026,
  author = {Mei, Xin and Yang, Libin and Gao, Dehong and Cai, Xiaoyan and Han, Junwei and Liu, Tianming},
  title = {FiR-Rad: Fine-Grained Reinforcement With Structured Reasoning for Chest X-Ray Report Generation},
  journal = {IEEE Transactions on Medical Imaging},
  year = {2026},
  volume = {45},
  number = {6},
  pages = {2922--2935},
  doi = {10.1109/tmi.2026.3665561}
}
Kostenlos starten →