DINOv3-MIL: Per-Kidney Multi-Label Tumour and Cyst Detection from Foundation-Model Patch Tokens on KiTS23
Quick Answer
This paper shows that DINOv3-MIL utilizes gated attention multiple instance learning for renal tumor and cyst detection, achieving AUROC scores of 0.74 and 0.80 respectively on the KiTS23 dataset.
Quick Take
This method outperforms traditional aggregators, enhancing interpretability while managing the trade-off in performance at scale.
Key Points
- Attention MIL achieves highest AUROC for tumors at 0.74 and cysts at 0.80.
- Prototype head shows poor transferability for cyst detection with AUROC of 0.51.
- Gated attention enhances attention scores by 7.5-9.8x over chance.
- Study involves 966 kidneys with 97 used for testing.
- Results accepted for presentation at MIUA 2026.
Paper Resources
📖 Reader Mode
~2 min readAbstract: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.
| Comments: | Accepted at MIUA 2026 (poster). To appear in Frontiers in Medical Technology |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV) |
| Cite as: | arXiv:2607.22687 [cs.CV] |
| (or arXiv:2607.22687v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2607.22687 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Sahil Sharma Dr. [view email]
[v1]
Thu, 16 Jul 2026 14:50:22 UTC (326 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CV
See more →ProMoE-FL: Prototype-conditioned Mixture of Experts for Multimodal Federated Learning with Missing Modalities
ProMoE-FL introduces a Prototype-conditioned Mixture-of-Experts framework for multimodal federated learning, effectively addressing missing modalities. It outperforms existing methods on four chest X-ray datasets, demonstrating superior feature synthesis capabilities in both homogeneous and heterogeneous settings.