# Are the Latent Representations of Foundation Models for Pathology Invariant to Rotation?

arXiv: https://arxiv.org/abs/2412.11938 · HTML: https://mlsanity.com/paper/2412.11938/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 16 Dec 2024.

## Built on by (most cited first, 4 of 4)

- [Scanner-Induced Domain Shifts Undermine the Robustness of Pathology Foundation Models](https://mlsanity.com/paper/2601.04163/) (2026)
- [Beyond the Failures: Rethinking Foundation Models in Pathology](https://mlsanity.com/paper/2510.23807/) (2025)
- [Controllable Latent Space Augmentation for Digital Pathology](https://mlsanity.com/paper/2508.14588/) (2025)
- [Aggregation Schemes for Single-Vector WSI Representation Learning in Digital Pathology](https://mlsanity.com/paper/2501.17822/) (2025)

## Abstract (arXiv)

Self-supervised foundation models for digital pathology encode small patches from H&E whole slide images into latent representations used for downstream tasks. However, the invariance of these representations to patch rotation remains unexplored. This study investigates the rotational invariance of latent representations across twelve foundation models by quantifying the alignment between non-rotated and rotated patches using mutual k-nearest neighbours and cosine distance. Models that incorporated rotation augmentation during self-supervised training exhibited significantly greater invariance to rotations. We hypothesise that the absence of rotational inductive bias in the transformer architecture necessitates rotation augmentation during training to achieve learned invariance. Code: https://github.com/MatousE/rot-invariance-analysis.