AEGIS联合建模乳腺癌筛查与密度评估，性能逼近人类专家水平。
AEGIS: A Multi-Task Joint-Embedding Predictive Architecture for Mammography
- 用自监督联合嵌入预训练+渐进分辨率微调，统一处理乳腺癌与密度分析
- 最大模型在785例测试集上癌症筛查达0.949 AUC，敏感性93%，特异性75%
- 对四种密度等级分类准确率达62.6%，跨人群零样本迁移表现稳健
我们提出AEGIS，一种用于乳腺癌检测与密度评估的联合嵌入预测架构。基于来自14个临床中心的71,103份乳腺钼靶数据，采用自监督联合嵌入预测架构（JEPA）预训练三种视觉变压器变体（Small/Base/Large），随后通过渐进式分辨率提升至2048x1536进行有监督微调。在精选的785例测试集上，最大模型在乳腺癌分诊任务中达到0.949 AUC，最优操作点下敏感性为93%，特异性为75%。结合美国食品药品管理局认证基线的集成模型进一步将判别能力提升至0.952 AUC。在乳腺密度分类任务中，二分类（致密 vs 非致密）的AUC为0.953，四类BI-RADS分类的精确度为62.6%，邻近类别准确率高达98.8%，接近人类阅片者间一致性水平。在公开的VinDr-Mammo数据集上进行外部验证，结果表明该模型在不同参考标准下的跨人群迁移能力良好，最大模型在零样本设置下的分诊任务实现0.871 AUC。
原文摘要 · Abstract (English)
We present Aegis, a joint-embedding predictive architecture for breast cancer detection and density assessment in mammography. We train three Vision Transformer variants (Small/Base/Large) using self-supervised joint-embedding predictive architecture (JEPA) pre-training on 71,103 studies from 14 clinical sites, followed by supervised fine-tuning with progressive resolution scaling up to 2048x1536. On a curated 785-study test set, our largest model achieves area under the receiver operating characteristic curve (AUC) 0.949 for breast cancer triage with 93% sensitivity and 75% specificity at the optimal operating point. An ensemble combining our model with a U.S. Food and Drug Administration-cleared baseline further improves discrimination to 0.952 AUC. For breast density classification, the model achieves 0.953 AUC for binary (dense vs. non-dense) classification and 62.6% exact accuracy across four Breast Imaging Reporting and Data System (BI-RADS) categories, with 98.8% adjacent accuracy comparable to reported human inter-reader agreement. External validation on the public VinDr-Mammo dataset provides evidence of cross-population transfer under a different reference standard, with the largest model achieving 0.871 AUC for triage in a zero-shot setting.
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。