Multi-Scale Representation of Follicular Lymphoma Pathology Images in a Single Hyperbolic Space

TL;DR

This paper introduces a novel self-supervised method to embed multi-scale lymphoma pathology images into a hyperbolic space, effectively capturing hierarchical tissue and cell structures for improved disease analysis.

Contribution

It presents a new approach for representing multi-scale pathology images in a single hyperbolic space, capturing hierarchical relationships and morphological changes during disease progression.

Findings

Effective encoding of hierarchical tissue and cell structures

Captures disease state and cell type variations

Utilizes Poincaré ball for hyperbolic embedding

Abstract

We propose a method for representing malignant lymphoma pathology images, from high-resolution cell nuclei to low-resolution tissue images, within a single hyperbolic space using self-supervised learning. To capture morphological changes that occur across scales during disease progression, our approach embeds tissue and corresponding nucleus images close to each other based on inclusion relationships. Using the Poincar\'e ball as the feature space enables effective encoding of this hierarchical structure. The learned representations capture both disease state and cell type variations.