Privileged Lesion-Context Relational Distillation for Mask-Free Skin Lesion Classification
作者: Abu Mukaddim Rahi, Md Mithun Hossain, Md Zulficar Hasan Joy, M. F. Mridha, Md. Jakir Hossen
分类: cs.CV
发布日期: 2026-07-21
备注: Manuscript submitted for consideration. The paper presents a mask-free skin lesion classification framework evaluated on HAM10000 and externally validated on ISIC 2018
💡 一句话要点
提出特权病变上下文关系蒸馏以解决无掩膜皮肤病变分类问题
🎯 匹配领域: 支柱二:RL算法与架构 (RL & Architecture) 支柱六:视频提取与匹配 (Video Extraction)
关键词: 皮肤病变分类 无掩膜推理 知识蒸馏 深度学习 医学影像分析 特权学习 上下文关系
📋 核心要点
- 现有方法在皮肤病变分类中依赖于分割掩膜,增加了临床应用的复杂性和计算负担。
- 论文提出的PLCRD框架在训练中利用病变掩膜,而在推理时仅依赖图像,提升了实用性。
- 实验结果显示,PLCRD在HAM10000上取得了0.773的宏F1分数,在ISIC 2018上也表现出色,验证了其有效性。
📝 摘要(中文)
准确的皮肤病变分类通常依赖于病变分割掩膜,但在推理过程中需要掩膜或辅助分割模型会降低临床实用性并增加计算复杂性。本文提出了特权病变上下文关系蒸馏(PLCRD),一种教师-学生框架,在训练过程中仅利用病变掩膜,同时保持图像推理的无掩膜特性。特权教师通过分析原始皮肤镜图像及其掩膜引导的病变区域,学习病变特异性和上下文诊断表示。然后,图像无掩膜的学生通过互补知识转移机制进行训练,传递教师的诊断分布、病变聚焦注意力、病变间关系几何和病变上下文结构。PLCRD在HAM10000数据集上进行了评估,并在ISIC 2018上进行了外部验证,未进行重新训练,取得了良好的效果。
🔬 方法详解
问题定义:本文旨在解决皮肤病变分类中对分割掩膜的依赖问题。现有方法在推理阶段需要掩膜或辅助模型,导致临床应用受限和计算复杂性增加。
核心思路:PLCRD框架通过教师-学生模型设计,在训练阶段利用病变掩膜,推理阶段则仅使用图像,从而实现无掩膜的分类。教师模型分析原始图像及其掩膜引导的区域,提取病变特征和上下文信息。
技术框架:PLCRD包括两个主要模块:特权教师和图像无掩膜的学生。教师模型通过掩膜学习病变特征,学生模型则通过知识转移机制获取教师的诊断信息和结构。
关键创新:PLCRD的创新在于通过关系蒸馏机制将教师的病变特征和上下文信息转移给学生,而不直接进行特征匹配,避免了异构架构间的直接对比。
关键设计:该方法采用了多种损失函数,包括病变间相似性对齐、病变上下文亲和性匹配和类感知关系学习,确保学生模型能够有效内化教师的诊断结构。
🖼️ 关键图片
📊 实验亮点
PLCRD在HAM10000数据集上实现了0.773的宏F1分数和0.764的平衡准确率,宏AUROC达到了0.976,显示出显著的分类性能。同时,在ISIC 2018上也取得了0.732的宏F1分数,验证了其有效性和可迁移性。
🎯 应用场景
该研究在皮肤病变分类领域具有广泛的应用潜力,尤其是在临床环境中,医生可以在没有复杂分割模型的情况下快速进行病变诊断。未来,该方法可能扩展到其他医学影像分析任务中,提升自动化诊断的效率和准确性。
📄 摘要(原文)
Accurate skin lesion classification can benefit from lesion segmentation masks, but requiring masks or an auxiliary segmentation model during inference reduces clinical practicality and increases computational complexity. This work introduces Privileged Lesion-Context Relational Distillation (PLCRD), a teacher-student framework that exploits lesion masks exclusively during training while preserving image-only inference. The privileged teacher jointly analyzes the original dermoscopic image and its mask-guided lesion region to learn lesion-specific and contextual diagnostic representations. An image-only student is then trained through complementary knowledge-transfer mechanisms that convey the teacher's diagnostic distribution, lesion-focused attention, inter-lesion relational geometry, and lesion-context structure. PLCRD decomposes deep representations into lesion and contextual embeddings and transfers their relational organization through inter-lesion similarity alignment, lesion-context affinity matching, separation regularization, and class-aware relational learning. This formulation avoids direct feature matching between heterogeneous teacher and student architectures and enables the student to internalize mask-informed diagnostic structure without accessing masks at deployment. The framework was evaluated on HAM10000 using lesion-disjoint data partitioning and externally validated on ISIC 2018 without retraining. PLCRD achieved a lesion-level macro-F1 of 0.773 +/- 0.018, balanced accuracy of 0.764 +/- 0.023, and macro-AUROC of 0.976 +/- 0.002 on HAM10000, together with a macro-F1 of 0.732 +/- 0.008 on ISIC 2018. The results indicate that privileged lesion annotations can be transformed into transferable relational knowledge, yielding a practical and interpretable approach to mask-free skin lesion classification.