用于乳腺X线摄影分类的基于扩散修复的健康反事实生成
Healthy Counterfactual Generation via Diffusion Inpainting for Mammography Classification
查看机构详情
- Institute for Systems and Robotics（系统与机器人研究所）
- Instituto Superior Técnico（里斯本高等理工学院）
机构由 AI 辅助整理，请以论文原文为准。
浏览论文内容
中文总结 AI 辅助
针对乳腺X线分类CAD系统假阴性问题，提出基于扩散修复的健康反事实生成策略，在四类模型上提升了固定80%特异性下的灵敏度，助力更可靠的乳腺癌CAD系统开发。
中文摘要 AI 辅助
乳腺癌筛查的计算机辅助诊断（CAD）系统存在假阴性这一关键局限，原因在于检测与治疗延迟。为解决该问题，本文提出一种反事实数据增强策略，通过从异常图像中“擦除”病灶来生成健康乳腺X线图像，从而丰富训练分布。我们在BI-RADS 1（健康）乳腺X线图像上训练去噪扩散概率模型（DDPM），并采用基于RePaint的采样策略，在标注的病灶边界框内修复出真实的正常组织。生成的健康反事实图像将标注的病灶区域替换为真实的健康组织，同时保留患者特有的解剖结构，真实图像与生成图像的相似度指标可验证这一点。放射科医生进一步评估了图像真实性，确认其与原始数据集质量一致。我们在四种代表性分类器架构上评估了反事实增强：卷积神经网络（ConvNeXt）、视觉Transformer（ViT）、基于乳腺X线图像-报告对预训练的视觉语言模型（Mammo-CLIP）以及基于多尺度注意力的多实例学习框架（FPN-MIL）。在VinDr-Mammo数据集上开展的实验显示，所有架构的灵敏度均有所提升，尤其是在固定特异性为80%时，这有助于开发更可靠的乳腺癌CAD系统。代码可在指定URL获取。
英文摘要
False negatives remain a critical limitation of computer-aided diagnosis (CAD) systems for breast cancer screening due to delayed detection and treatment. To address this issue, we propose a counterfactual data augmentation strategy that generates healthy mammograms by "erasing" lesions from anomalous images, thereby enriching the training distribution. We train a Denoising Diffusion Probabilistic Model on BI-RADS 1 (healthy) mammograms and use a RePaint-based sampling strategy to inpaint realistic normal tissue within annotated lesion bounding boxes. The resulting healthy counterfactuals replace annotated lesion regions with realistic healthy tissue while preserving patient-specific anatomical structure, as supported by similarity metrics between real and generated images. Image realism was further assessed by radiologists and found to be consistent with the original dataset quality. We evaluate counterfactual augmentation across four representative classifier architectures: a convolutional neural network (ConvNeXt), a vision transformer (ViT), a vision-language model pre-trained on mammogram-report pairs (Mammo-CLIP) and a multi-scale attention-based multiple-instance learning framework (FPN-MIL). Experiments conducted on the VinDr-Mammo dataset show improvements in sensitivity across all architectures, particularly at 80\% fixed specificity, contributing towards more reliable CAD systems for breast cancer. Code is available at: https://github.com/ines03garcia/diffusion-based-counterfactual-generation.
