# Learning Where to Focus: Self-Supervised Multi-Scale ViTs for Histopathology

arXiv: https://arxiv.org/abs/2609.18578 · HTML: https://mlsanity.com/paper/2609.18578/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 37 of its 60 references that are in the map.
- First posted on arXiv 16 Sep 2026.

## Builds on (most cited first, 30 of 37)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [DINOv2: Learning Robust Visual Features without Supervision](https://mlsanity.com/paper/2304.07193/) (2023)
- [Representation Learning with Contrastive Predictive Coding](https://mlsanity.com/paper/1807.03748/) (2018)
- [Emerging Properties in Self-Supervised Vision Transformers](https://mlsanity.com/paper/2104.14294/) (2021)
- [Barlow Twins: Self-Supervised Learning via Redundancy Reduction](https://mlsanity.com/paper/2103.03230/) (2021)
- [An Empirical Study of Training Self-Supervised Vision Transformers](https://mlsanity.com/paper/2104.02057/) (2021)
- [Understanding Contrastive Representation Learning through Alignment and Uniformity on the Hypersphere](https://mlsanity.com/paper/2005.10242/) (2020)
- [Attention-based Deep Multiple Instance Learning](https://mlsanity.com/paper/1802.04712/) (2018)
- [Token Merging: Your ViT But Faster](https://mlsanity.com/paper/2210.09461/) (2022)
- [How to train your ViT? Data, Augmentation, and Regularization in Vision Transformers](https://mlsanity.com/paper/2106.10270/) (2021)
- [Dual-stream Multiple Instance Learning Network for Whole Slide Image Classification with Self-supervised Contrastive Learning](https://mlsanity.com/paper/2011.08939/) (2020)
- [Not All Patches are What You Need: Expediting Vision Transformers via Token Reorganizations](https://mlsanity.com/paper/2202.07800/) (2022)
- [Scaling Vision Transformers to Gigapixel Images via Hierarchical Self-Supervised Learning](https://mlsanity.com/paper/2206.02647/) (2022)
- [Adaptive Token Sampling For Efficient Vision Transformers](https://mlsanity.com/paper/2111.15667/) (2021)
- [SPViT: Enabling Faster Vision Transformers via Soft Token Pruning](https://mlsanity.com/paper/2112.13890/) (2021)
- [Benchmarking Self-Supervised Learning on Diverse Pathology Datasets](https://mlsanity.com/paper/2212.04690/) (2022)
- [Self supervised contrastive learning for digital histopathology](https://mlsanity.com/paper/2011.13971/) (2020)
- [QuadTree Attention for Vision Transformers](https://mlsanity.com/paper/2201.02767/) (2022)
- [Virchow: A Million-Slide Digital Pathology Foundation Model](https://mlsanity.com/paper/2309.07778/) (2023)
- [CF-ViT: A General Coarse-to-Fine Method for Vision Transformer](https://mlsanity.com/paper/2203.03821/) (2022)
- [Self-Supervised Vision Transformers Learn Visual Concepts in Histopathology](https://mlsanity.com/paper/2203.00585/) (2022)
- [Differentiable Zooming for Multiple Instance Learning on Whole-Slide Images](https://mlsanity.com/paper/2204.12454/) (2022)
- [A General-Purpose Self-Supervised Model for Computational Pathology](https://mlsanity.com/paper/2308.15474/) (2023)
- [Distilling foundation models for robust and efficient models in digital pathology](https://mlsanity.com/paper/2501.16239/) (2025)
- [Computational Pathology at Health System Scale -- Self-Supervised Foundation Models from Three Billion Images](https://mlsanity.com/paper/2310.07033/) (2023)
- [ReMix: A General and Efficient Framework for Multiple Instance Learning based Whole Slide Image Classification](https://mlsanity.com/paper/2207.01805/) (2022)
- [Self-slimmed Vision Transformer](https://mlsanity.com/paper/2111.12624/) (2021)
- [ScoreNet: Learning Non-Uniform Attention and Augmentation for Transformer-Based Histopathological Image Classification](https://mlsanity.com/paper/2202.07570/) (2022)
- [MSViT: Dynamic Mixed-Scale Tokenization for Vision Transformers](https://mlsanity.com/paper/2307.02321/) (2023)

## Abstract (arXiv)

Pathologists diagnose diseases by first locating suspicious tissue and then examining it at higher magnification, whereas self-supervised vision transformers (ViTs) allocate the same spatial resolution to every image region despite diagnostic evidence being sparse and spanning multiple biological scales. Recent pathology foundation models have substantially improved representation quality by scaling training data and model capacity, but largely retain uniform tokenization. We instead investigate whether pathology representations can be improved by learning where to allocate spatial resolution during self-supervised learning. To this end, we propose CRAFT (Coarse-to-fine Region-Adaptive Feature Tokenization), a DINO-based framework that learns image-dependent mixed-scale representations by using self-supervised attention to selectively refine informative regions while preserving coarse context, together with a symmetric cross-scale regularization objective that encourages complementary coarse and fine representations. Across CAMELYON16, TCGA-Lung subtype classification, and TCGA-LUAD survival prediction, CRAFT consistently outperforms comparable-scale self-supervised methods while requiring lower inference computation. Despite using only a compact 22M parameter backbone trained on comparatively small pathology datasets, CRAFT remains competitive with, and often surpasses, substantially larger pathology foundation models.