arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2601.09136cs.CVcs.AI
收藏

SkinFlow: 通过动态视觉编码和分阶段强化学习实现开放性皮肤病诊断的高效信息传输

SkinFlow: Efficient Information Transmission for Open Dermatological Diagnosis via Dynamic Visual Encoding and Staged RL

Baichuan Inc.(百川公司);Department of Dermatology Peking University First Hospital(北京大学第一医院皮肤科);Beijing Key Laboratory of Molecular Diagnosis on Dermatoses(北京分子皮肤病诊断重点实验室);National Clinical Research Center for Skin and Sexually Transmitted Diseases(国家皮肤及性传播疾病临床医学研究中心);NMPA Key Laboratory for Quality Control and Evaluation of Cosmetics(国家药监局化妆品质量控制与评价重点实验室);School of Biomedical Engineering Tsinghua University(清华大学生物医学工程学院);University of Hong Kong(香港大学)

Lijun Liu, Linwei Chen, Zhishou Zhang, Meng Tian, Hengfu Cui, Ruiyang Li, Zhaocheng Liu, Qiang Ju, Qianxi Li, Hong-Yu Zhou

AI总结:

SkinFlow通过动态视觉编码和分阶段强化学习提升皮肤病诊断效率,实现比通用模型更优的准确率

摘要:

通用的大规模视觉-语言模型(LVLMs)尽管规模庞大,但在皮肤病学中往往因“扩散注意力”而表现不佳——即无法将细微的病理病变与背景噪声区分开来。在本文中,我们挑战了参数扩展是实现医学精度的唯一途径的假设。我们引入了SkinFlow框架,将诊断视为视觉信息传输效率的优化问题。我们的方法利用虚拟宽度动态视觉编码器(DVE)在不进行物理参数扩展的情况下“展开”复杂的病理流形,并结合两阶段强化学习策略。该策略依次对齐显性医学描述(第一阶段)并在受限语义空间中重建隐性诊断纹理(第二阶段)。此外,我们提出了一种以临床为导向的评估协议,优先考虑诊断安全性和层次相关性而非刚性标签匹配。实证结果令人印象深刻:我们的7B模型在Fitzpatrick17k基准上建立了新的状态-of-the-art,相较于大规模通用模型(如Qwen3VL-235B和GPT-5.2)在Top-1准确率上提高了+12.06%,在Top-6准确率上提高了+28.57%。这些发现表明,优化几何容量和信息流比单纯增加参数规模能产生更优越的诊断推理效果。

英文摘要:

General-purpose Large Vision-Language Models (LVLMs), despite their massive scale, often falter in dermatology due to "diffuse attention" - the inability to disentangle subtle pathological lesions from background noise. In this paper, we challenge the assumption that parameter scaling is the only path to medical precision. We introduce SkinFlow, a framework that treats diagnosis as an optimization of visual information transmission efficiency. Our approach utilizes a Virtual-Width Dynamic Vision Encoder (DVE) to "unfold" complex pathological manifolds without physical parameter expansion, coupled with a two-stage Reinforcement Learning strategy. This strategy sequentially aligns explicit medical descriptions (Stage I) and reconstructs implicit diagnostic textures (Stage II) within a constrained semantic space. Furthermore, we propose a clinically grounded evaluation protocol that prioritizes diagnostic safety and hierarchical relevance over rigid label matching. Empirical results are compelling: our 7B model establishes a new state-of-the-art on the Fitzpatrick17k benchmark, achieving a +12.06% gain in Top-1 accuracy and a +28.57% boost in Top-6 accuracy over the massive general-purpose models (e.g., Qwen3VL-235B and GPT-5.2). These findings demonstrate that optimizing geometric capacity and information flow yields superior diagnostic reasoning compared to raw parameter scaling.

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。

↑