# A Distributional Robustness Margin For Pathology Foundation Models

arXiv: https://arxiv.org/abs/2607.25497 · HTML: https://mlsanity.com/paper/2607.25497/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 20 of its 41 references that are in the map.
- First posted on arXiv 28 Jul 2026.

## Builds on (most cited first, 20 of 20)

- [DINOv2: Learning Robust Visual Features without Supervision](https://mlsanity.com/paper/2304.07193/) (2023)
- [Emerging Properties in Self-Supervised Vision Transformers](https://mlsanity.com/paper/2104.14294/) (2021)
- [Shortcut Learning in Deep Neural Networks](https://mlsanity.com/paper/2004.07780/) (2020)
- [Self-supervised Visual Feature Learning with Deep Neural Networks: A Survey](https://mlsanity.com/paper/1902.06162/) (2019)
- [Quantifying the effects of data augmentation and stain color normalization in convolutional neural networks for computational pathology](https://mlsanity.com/paper/1902.06543/) (2019)
- [Multimodal Whole Slide Foundation Model for Pathology](https://mlsanity.com/paper/2411.19666/) (2024)
- [Virchow2: Scaling Self-Supervised Mixed Magnification Models in Pathology](https://mlsanity.com/paper/2408.00738/) (2024)
- [PRISM: A Multi-Modal Generative Foundation Model for Slide-Level Histopathology](https://mlsanity.com/paper/2405.10254/) (2024)
- [Phikon-v2, A large and public feature extractor for biomarker prediction](https://mlsanity.com/paper/2409.09173/) (2024)
- [A Clinical Benchmark of Public Self-Supervised Pathology Foundation Models](https://mlsanity.com/paper/2407.06508/) (2024)
- [A Multimodal Knowledge-enhanced Whole-slide Pathology Foundation Model](https://mlsanity.com/paper/2407.15362/) (2024)
- [Hibou: A Family of Foundational Vision Transformers for Pathology](https://mlsanity.com/paper/2406.05074/) (2024)
- [Benchmarking foundation models as feature extractors for weakly-supervised computational pathology](https://mlsanity.com/paper/2408.15823/) (2024)
- [Current Pathology Foundation Models are unrobust to Medical Center Differences](https://mlsanity.com/paper/2501.18055/) (2025)
- [Distilling foundation models for robust and efficient models in digital pathology](https://mlsanity.com/paper/2501.16239/) (2025)
- [Training state-of-the-art pathology foundation models with orders of magnitude less data](https://mlsanity.com/paper/2504.05186/) (2025)
- [Do Histopathological Foundation Models Eliminate Batch Effects? A Comparative Study](https://mlsanity.com/paper/2411.05489/) (2024)
- [The PAR dataset: Prostate biopsy whole slide images from an underrepresented Middle Eastern population](https://mlsanity.com/paper/2512.03854/) (2025)
- [Deep Learning From Routine Histology Improves Risk Stratification for Biochemical Recurrence in Prostate Cancer](https://mlsanity.com/paper/2603.14187/) (2026)
- [MOOZY: A Patient-First Foundation Model for Computational Pathology](https://mlsanity.com/paper/2603.27048/) (2026)

## Abstract (arXiv)

Pathology foundation models encode non-biological variation introduced by tissue preparation, staining and scanning, enabling shortcut learning that undermines generalisation across institutions. The Robustness Index (RI) was proposed to assess whether local representation geometry is dominated by biological or non-biological variation. However, its construction suffers from structural limitations that make cross-model comparison unreliable, calling for a more principled metric. We introduce the Cross-confounder Robustness Margin (CRoMa), a signed, per-sample margin that measures whether samples sharing the same biology but different confounder lie closer than samples sharing the same confounder but different biology. It is defined for every sample, allowing models to be compared on the same cohort and robustness to be analysed as a distribution rather than reduced to a single pooled score. We evaluated CRoMa across 20 tile-level encoders on three benchmarks. Rankings by median CRoMa were highly consistent across benchmarks (Spearman rho ~ 0.90), yet every encoder retained confounder-dominated samples, whose prevalence and severity varied markedly. Similar patterns emerged for four slide-level encoders evaluated on a separate benchmark, extending the analysis beyond tile-level representations. Higher median CRoMa was associated with smaller shortcut-induced performance losses in downstream linear probes, supporting its use as a representation-level indicator of shortcut susceptibility.