用视觉大模型分析肾脏影像，精准识别肿瘤和囊肿。
DINOv3-MIL: Per-Kidney Multi-Label Tumour and Cyst Detection from Foundation-Model Patch Tokens on KiTS23
- 用注意力机制的多实例学习聚合海量图像块特征
- 肿瘤检测AUROC达0.74，囊肿达0.80，显著优于其他方法
- 适合医学影像分析、需要可解释性的研究者使用
在未进行领域预训练的情况下，基于自然图像训练的视觉基础模型可直接迁移至医疗任务。然而，对于体积分类，需整合每例研究中数万张图像块特征，而聚合器限制了模型的可解释性。我们在相同的冻结DINOv3 ViT-H/16+特征上对比三种聚合器：CLS-token线性探测、基于55,296个图像块的门控注意力多实例学习（MIL），以及遵循ProtoViT的原型头。注意力MIL在肾肿瘤（AUROC 0.74，95%置信区间0.64–0.83）和囊肿（AUROC 0.80，0.70–0.88）检测上表现最佳，且在标注病灶内注意力增强程度达随机水平的7.5–9.8倍。原型头在囊肿检测上未能迁移（AUROC 0.51），揭示了在此图像块规模下可解释性与性能之间的权衡。
原文摘要 · Abstract (English)
Foundation vision models trained on natural images transfer to medical tasks without domain pre-training, but volumetric classification requires aggregating tens of thousands of patch tokens per study, and the aggregator constrains how the resulting model can be interpreted. We compare three aggregators on identical frozen DINOv3 ViT-H/16+ features for renal tumour/cyst detection on KiTS23 (966 kidneys; n=97 test): a CLS-token linear probe, gated attention multiple instance learning (MIL) over 55,296 patch tokens, and a prototype head following ProtoViT. Attention MIL achieves the highest AUROC for tumour (0.74, 95% CI 0.64-0.83) and cyst (0.80, 0.70-0.88), with attention enriched 7.5-9.8x over chance within annotated lesions. The prototype head does not transfer to cyst detection (AUROC 0.51), exposing an interpretability-performance trade-off at this token scale.
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。