X-Ragent improves chest X-ray analysis by selecting valid evidence
X-Ragent: Tool-Augmented Reinforcement Learning for Evidence-Grounded Chest X-Ray Reasoning
Computer Vision and Pattern Recognition
Summary
Chest X-rays are hard to interpret because doctors need to combine the images with specific measurements and evidence. The authors created X-Ragent, a method that helps an AI agent decide when to ask for extra information and only accept valid measurements to make better diagnoses. They built a training dataset with examples of reasoning and tested this method on three benchmarks, where it performed better than previous approaches. This makes the AI's decisions more reliable and based on sound evidence.
What this means in practice
- •For clinical radiology teams: Assist radiologists by providing AI-based suggestions grounded in verified evidence for more accurate chest X-ray diagnoses.
- •For medical imaging software developers: Incorporate X-Ragent’s evidence-selection framework to improve diagnostic AI tools that integrate image analysis with measurement data.$Commercial implications: Enables development of advanced diagnostic software that offers trustworthy X-ray analysis with evidence validation for hospitals and clinics.
Authors
Yi Wang, Zhenlong Yuan, Yuchan Guo, Maolei Huang, Zijian Song, Dapeng Zhang, Rongbin Tan, Jiyuan Wang, Zhiyuan Wang, Fangfang Lin, Shuhan Song, Huawei Cao
Abstract
Chest X-ray (CXR) interpretation often requires integrating visual findings with task-specific evidence rather than relying on direct visual recognition alone. While tool-augmented medical agents can supply such evidence via external modules, they often lack explicit mechanisms for learning when auxiliary evidence is needed and for preventing inapplicable quantitative evidence from entering diagnostic reasoning. To address this, we propose X-Ragent, a post-training framework for tool-augmented reinforcement learning that decouples learned auxiliary-evidence seeking from validity-gated quantitative measurement acquisition for evidence-grounded CXR reasoning. We construct Rad-CoT, a 7.7K-sample corpus combining causally ordered tool-augmented trajectories with report-style diagnostic reasoning, and use it to initialize Qwen3-VL-8B via agentic supervised fine-tuning (SFT). We then optimize the policy with Group Relative Policy Optimization (GRPO) using an accuracy-gated reward that applies process-level terms for measurement consistency and auxiliary-tool cost only to correct trajectories, preventing plausible process behavior from compensating for an incorrect diagnosis. Across three CXR benchmarks, X-Ragent achieves the strongest results, including an 8.6-point improvement on the fully held-out ChestAgentBench. By seeking evidence selectively and admitting only valid measurements, X-Ragent supports more faithful and evidence-grounded CXR reasoning.