Think Twice to See More：医疗 VLMs 中的迭代视觉推理
Think Twice to See More: Iterative Visual Reasoning in Medical VLMs
AI总结：
针对医疗 VLM 单次推理忽略局部视觉线索的问题，提出 ViTAR 框架，通过“思考-行动-再思考-回答”认知链模拟专家迭代推理，结合监督微调与强化学习策略，提升了医疗 AI 的性能与可信度。
摘要：
医疗视觉语言模型（VLMs）擅长图文理解，但通常依赖单次推理，忽略了局部视觉线索。然而在临床实践中，人类专家在得出最终诊断前会迭代地扫描、聚焦并细化感兴趣区域。为缩小这一机器与人类感知的差距，我们引入了 ViTAR，一种新型 VLM 框架，通过“思考-行动-再思考-回答”的认知链来模拟人类专家的迭代推理过程。ViTAR 将医疗图像视为交互对象，使模型能够进行多步视觉推理。为支持该方法，我们构建了包含 1K 个交互样本的高质量指令数据集，编码了专家级的诊断行为。此外，还整理了 16K 面向细粒度视觉诊断的视觉问答训练数据。我们引入了两阶段训练策略：首先进行监督微调以引导认知轨迹，随后采用强化学习优化决策。广泛评估表明，ViTAR 优于强大的最先进模型。视觉注意力分析显示，从“思考”到“再思考”轮次，ViTAR 越来越多地将视觉定位锚定在临床关键区域，并在推理期间保持对视觉 token 的高注意力分配，为其性能提升提供了机制性见解。这些发现表明，将专家风格的迭代思维链嵌入 VLMs 能提升医疗 AI 的性能与可信度。
英文摘要：
Medical vision-language models (VLMs) excel at image-text understanding but typically rely on a single-pass reasoning that neglects localized visual cues. In clinical practice, however, human experts iteratively scan, focus, and refine the regions of interest before reaching a final diagnosis. To narrow this machine-human perception gap, we introduce ViTAR, a novel VLM framework that emulates the iterative reasoning process of human experts through a cognitive chain of "think-act-rethink-answer". ViTAR treats medical images as interactive objects, enabling models to engage multi-step visual reasoning. To support this approach, we curate a high-quality instruction dataset comprising 1K interactive examples that encode expert-like diagnostic behaviors. In addition, a 16K visual question answering training data has been curated towards fine-grained visual diagnosis. We introduce a two-stage training strategy that begins with supervised fine-tuning to guide cognitive trajectories, followed by the reinforcement learning to optimize decision-making. Extensive evaluations demonstrate that ViTAR outperforms strong state-of-the-art models. Visual attention analysis reveals that from the "think" to "rethink" rounds, ViTAR increasingly anchors visual grounding to clinically critical regions and maintains high attention allocation to visual tokens during reasoning, providing mechanistic insight into its improved performance. These findings demonstrate that embedding expert-style iterative thinking chains into VLMs enhances both performance and trustworthiness of medical AI.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
