评估医学影像中视觉Transformer的可解释性
Evaluating the Explainability of Vision Transformers in Medical Imaging
AI总结：
本研究评估ViT、DeiT、DINO和Swin Transformer在医学影像中的可解释性，发现DINO结合Grad-CAM提供最忠实和局部化的解释，支持ViTs在临床诊断中的可靠集成。
摘要：
理解模型决策在医学影像中至关重要，因为可解释性直接影响临床信任和采用。视觉Transformer（ViTs）在诊断影像中已展现出最先进的性能；然而，其复杂的注意力机制对可解释性构成挑战。本研究使用梯度注意力展开（Gradient Attention Rollout）和Grad-CAM，评估不同视觉Transformer架构和预训练策略——ViT、DeiT、DINO和Swin Transformer——的可解释性。我们在两项医学影像任务上进行定量和定性分析：外周血细胞分类和乳腺超声图像分类。我们的研究结果表明，DINO与Grad-CAM结合在数据集上提供了最忠实和局部化的解释。Grad-CAM始终生成类别判别性强且空间精确的热力图，而梯度注意力展开则产生更分散的激活。即使在错误分类的情况下，DINO与Grad-CAM也会突出显示似乎误导了模型的临床相关形态特征。通过提高模型透明度，本研究支持将ViTs可靠且可解释地集成到关键医学诊断工作流程中。
英文摘要：
Understanding model decisions is crucial in medical imaging, where interpretability directly impacts clinical trust and adoption. Vision Transformers (ViTs) have demonstrated state-of-the-art performance in diagnostic imaging; however, their complex attention mechanisms pose challenges to explainability. This study evaluates the explainability of different Vision Transformer architectures and pre-training strategies - ViT, DeiT, DINO, and Swin Transformer - using Gradient Attention Rollout and Grad-CAM. We conduct both quantitative and qualitative analyses on two medical imaging tasks: peripheral blood cell classification and breast ultrasound image classification. Our findings indicate that DINO combined with Grad-CAM offers the most faithful and localized explanations across datasets. Grad-CAM consistently produces class-discriminative and spatially precise heatmaps, while Gradient Attention Rollout yields more scattered activations. Even in misclassification cases, DINO with Grad-CAM highlights clinically relevant morphological features that appear to have misled the model. By improving model transparency, this research supports the reliable and explainable integration of ViTs into critical medical diagnostic workflows.
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。
