基于视觉基础模型的可解释糖尿病视网膜病变分类
Explainable Diabetic Retinopathy Classification Using Vision Foundation Models
浏览论文内容
中文总结 AI 辅助
本研究提出基于视觉基础模型的可解释DR分类框架，评估DINOv2等模型及LoRA等训练策略，在ODIR、APTOS等数据集上取得良好分类性能，验证了模型注意力与临床病灶的相关性。
中文摘要 AI 辅助
糖尿病视网膜病变（DR）是可预防失明的主要原因，因此需要准确且可信的自动筛查方法。本研究探索一种基于视觉基础模型（vision foundation models）和多种迁移学习策略的可解释DR分类框架。评估了三种骨干网络：DINOv2、CLIP和Vision Transformer（ViT），采用全微调（full fine-tuning）、线性探测（linear probing）和低秩适配（Low-Rank Adaptation, LoRA）三种训练方式。模型在ODIR数据集上进行内部训练与评估，在APTOS数据集上进行外部评估以测试泛化能力。DINOv2-LoRA取得最高内部受试者工作特征曲线下面积（AUROC）0.758，而DINOv2全微调与ViT全微调取得最高外部AUROC 0.920。经等渗回归后，通过可靠性分析进一步评估校准度。可解释性方面，采用Grad-CAM和HiResCAM，结合IDRiD数据集的专家标注病灶掩码，用戴斯系数（Dice）、交并比（IoU）和指向游戏（Pointing Game）指标评估。结果表明，基础模型（尤其是DINOv2）可提供强预测性能，LoRA是全微调的参数高效替代方案；对解释图的定量评估进一步支持判断模型注意力是否对应临床相关视网膜病灶。
英文摘要
Diabetic retinopathy (DR) is a major cause of preventable blindness, creating a need for accurate and trustworthy automated screening. This study investigates an explainable DR classification framework using vision foundation models and multiple transfer learning strategies. Three backbones, DINOv2, CLIP, and Vision Transformer (ViT), were evaluated using full fine-tuning, linear probing, and Low-Rank Adaptation (LoRA). Models were trained and internally evaluated on the ODIR dataset and externally evaluated on APTOS to assess generalization. DINOv2-LoRA achieved the highest internal AUROC of 0.758, while DINOv2 full fine-tuning and ViT full fine-tuning achieved the highest external AUROC of 0.920. Calibration was further assessed using reliability analysis after isotonic regression. For explainability, Grad-CAM and HiResCAM were evaluated against expert-annotated lesion masks from the IDRiD dataset using Dice, Intersection over Union (IoU), and Pointing Game metrics. The results demonstrate that foundation models, particularly DINOv2, can provide strong predictive performance, while LoRA offers a parameter-efficient alternative to full fine-tuning. Quantitative evaluation of explanation maps further supports the assessment of whether model attention corresponds to clinically relevant retinal lesions.
发表机构
- Carl von Ossietzky Universität Oldenburg（卡尔·冯·奥西茨基奥尔登堡大学）
机构由 AI 辅助整理，请以论文原文为准。
