通过思维图遍历提升胸片报告生成的逻辑性与准确性
Thought Graph Traversal for Test-time Scaling in Chest X-ray VLLMs
- 构建医学先验引导的思维图，按器官顺序推理
- 动态扩展生成过程，实现测试时推理深度可控
- 无需训练即可提升报告一致性，适合医疗场景应用
测试时缩放为视觉语言大模型（VLLMs）提供了一种无需额外训练即可提升推理性能的前景广阔的方法。本文探索了一种简单而有效的方法，将测试时缩放应用于胸片报告生成。具体而言，我们提出了轻量级的思维图遍历（TGT）框架，通过结构化医学先验引导模型按医学上连贯的顺序推理器官特异性发现。该框架将医学先验融入提示，使模型在不改变底层模型的情况下进行更深入、更合逻辑的分析。为进一步增强推理深度，我们采用推理预算强制策略，在测试时通过动态扩展生成过程调整模型的推理深度。这一简单而强大的组合使冻结的放射科VLLM能够自我修正，生成更准确、一致的胸片报告。我们的方法在标准基准上优于基线提示方法，并能通过可追溯的推理路径揭示数据集偏差。代码与提示已开源，详见 https://github.com/glerium/Thought-Graph-Traversal。
原文摘要 · Abstract (English)
Test-time scaling offers a promising way to improve the reasoning performance of vision-language large models (VLLMs) without additional training. In this paper, we explore a simple but effective approach for applying test-time scaling to chest X-ray report generation. Specifically, we introduce a lightweight Thought Graph Traversal (TGT) framework that guides the model to reason through organ-specific findings in a medically coherent order. This framework integrates structured medical priors into the prompt, enabling deeper and more logical analysis with no changes to the underlying model. To further enhance reasoning depth, we apply a reasoning budget forcing strategy that adjusts the model's inference depth at test time by dynamically extending its generation process. This simple yet powerful combination allows a frozen radiology VLLM to self-correct and generate more accurate, consistent chest X-ray reports. Our method outperforms baseline prompting approaches on standard benchmarks, and also reveals dataset biases through traceable reasoning paths. Code and prompts are open-sourced for reproducibility at https://github.com/glerium/Thought-Graph-Traversal
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。