Saved in:
| Main Authors: | , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2506.12849 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866913895311998976 |
|---|---|
| author | Jiang, Songtao Wang, Yuan Chen, Ruizhe Zhang, Yan Luo, Ruilin Lei, Bohan Song, Sibo Feng, Yang Sun, Jimeng Wu, Jian Liu, Zuozhu |
| author_facet | Jiang, Songtao Wang, Yuan Chen, Ruizhe Zhang, Yan Luo, Ruilin Lei, Bohan Song, Sibo Feng, Yang Sun, Jimeng Wu, Jian Liu, Zuozhu |
| contents | In medical visual question answering (Med-VQA), achieving accurate responses relies on three critical steps: precise perception of medical imaging data, logical reasoning grounded in visual input and textual questions, and coherent answer derivation from the reasoning process. Recent advances in general vision-language models (VLMs) show that large-scale reinforcement learning (RL) could significantly enhance both reasoning capabilities and overall model performance. However, their application in medical domains is hindered by two fundamental challenges: 1) misalignment between perceptual understanding and reasoning stages, and 2) inconsistency between reasoning pathways and answer generation, both compounded by the scarcity of high-quality medical datasets for effective large-scale RL. In this paper, we first introduce Med-Zero-17K, a curated dataset for pure RL-based training, encompassing over 30 medical image modalities and 24 clinical tasks. Moreover, we propose a novel large-scale RL framework for Med-VLMs, Consistency-Aware Preference Optimization (CAPO), which integrates rewards to ensure fidelity between perception and reasoning, consistency in reasoning-to-answer derivation, and rule-based accuracy for final responses. Extensive experiments on both in-domain and out-of-domain scenarios demonstrate the superiority of our method over strong VLM baselines, showcasing strong generalization capability to 3D Med-VQA benchmarks and R1-like training paradigms. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_12849 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | CAPO: Reinforcing Consistent Reasoning in Medical Decision-Making Jiang, Songtao Wang, Yuan Chen, Ruizhe Zhang, Yan Luo, Ruilin Lei, Bohan Song, Sibo Feng, Yang Sun, Jimeng Wu, Jian Liu, Zuozhu Computer Vision and Pattern Recognition In medical visual question answering (Med-VQA), achieving accurate responses relies on three critical steps: precise perception of medical imaging data, logical reasoning grounded in visual input and textual questions, and coherent answer derivation from the reasoning process. Recent advances in general vision-language models (VLMs) show that large-scale reinforcement learning (RL) could significantly enhance both reasoning capabilities and overall model performance. However, their application in medical domains is hindered by two fundamental challenges: 1) misalignment between perceptual understanding and reasoning stages, and 2) inconsistency between reasoning pathways and answer generation, both compounded by the scarcity of high-quality medical datasets for effective large-scale RL. In this paper, we first introduce Med-Zero-17K, a curated dataset for pure RL-based training, encompassing over 30 medical image modalities and 24 clinical tasks. Moreover, we propose a novel large-scale RL framework for Med-VLMs, Consistency-Aware Preference Optimization (CAPO), which integrates rewards to ensure fidelity between perception and reasoning, consistency in reasoning-to-answer derivation, and rule-based accuracy for final responses. Extensive experiments on both in-domain and out-of-domain scenarios demonstrate the superiority of our method over strong VLM baselines, showcasing strong generalization capability to 3D Med-VQA benchmarks and R1-like training paradigms. |
| title | CAPO: Reinforcing Consistent Reasoning in Medical Decision-Making |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2506.12849 |