提出新方法提升乳腺钼靶图像跨机构泛化能力
BreastMammo and DenseMammo: Benchmarks for Mammography Domain Generalization
- 用前景直方图匹配解决不同医院设备带来的图像差异
- 在多个数据集上实现98.32%最高AUC，显著优于现有方法
- 适合医疗AI研究者用于提升模型跨机构稳定性
乳腺密度分类是乳腺癌风险评估的关键环节，但人工智能模型常因不同医疗机构的设备采集风格差异而难以泛化。本文引入两个新数据集BreastMammo和DenseMammo，以支持多视角乳腺钼靶图像研究。提出一种领域泛化框架，采用仅对前景区域进行直方图匹配的方法，缓解由不同临床来源引起的分布偏移问题。内部评估采用5折交叉验证，基于Swin Transformer骨干网络的模型在密度分类任务中达到最高98.32%的AUC。外部评估在TNMammo和LUMINA数据集上显示，该方法持续降低领域偏移，显著优于MixStyle及基于离散傅里叶变换的主流领域泛化方法。
原文摘要 · Abstract (English)
Breast density classification is a critical component of breast cancer risk assessment, yet AI models often struggle to generalize across clinical sites due to vendor-specific acquisition styles. In this work, we introduce two new datasets, BreastMammo and DenseMammo, to facilitate robust multi-view mammography research. We propose a domain generalization framework that utilizes a foreground-only histogram matching protocol to resolve the domain shift issue arising from disparate clinical sources. Internal evaluation using a 5-fold cross-validation protocol demonstrates the efficacy of our approach, with the Swin Transformer backbone achieving a peak AUC of 98.32% for density classification. External evaluation on the TNMammo and LUMINA datasets demonstrates that the proposed approach consistently reduces domain shift, significantly outperforming prominent domain generalization paradigms, including MixStyle and Discrete-Fourier-Transform-based frameworks.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。