ALBRT: Cellular Composition Prediction in Routine Histology Images

TL;DR

ALBRT is a novel image-based method that uses contrastive learning to accurately predict cellular composition in histology images, aiding various computational pathology tasks.

Contribution

It introduces a contrastive-learning inspired model that learns rotation-invariant features for cell prediction, outperforming existing methods and enabling transferability to new datasets.

Findings

Significant improvement over state-of-the-art in cell classification and counting

Features are transferable to new datasets for cellular analysis

Code and webserver are publicly available

Abstract

Cellular composition prediction, i.e., predicting the presence and counts of different types of cells in the tumor microenvironment from a digitized image of a Hematoxylin and Eosin (H&E) stained tissue section can be used for various tasks in computational pathology such as the analysis of cellular topology and interactions, subtype prediction, survival analysis, etc. In this work, we propose an image-based cellular composition predictor (ALBRT) which can accurately predict the presence and counts of different types of cells in a given image patch. ALBRT, by its contrastive-learning inspired design, learns a compact and rotation-invariant feature representation that is then used for cellular composition prediction of different cell types. It offers significant improvement over existing state-of-the-art approaches for cell classification and counting. The patch-level feature representation learned by ALBRT is transferrable for cellular composition analysis over novel datasets and can also be utilized for downstream prediction tasks in CPath as well. The code and the inference webserver for the proposed method are available at the URL: https://github.com/engrodawood/ALBRT.