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

TL;DR

This paper investigates whether foundation models for digital pathology produce rotation-invariant representations, finding that models trained with rotation augmentation exhibit greater invariance, highlighting the importance of training strategies.

Contribution

The study systematically evaluates rotational invariance in foundation models for pathology and demonstrates the impact of rotation augmentation during training.

Findings

Models with rotation augmentation show higher invariance.

Transformer architectures lack inherent rotational invariance.

Rotation augmentation improves downstream task robustness.

Abstract

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.