Summary of Point Transformer with Federated Learning for Predicting Breast Cancer HER2 Status from Hematoxylin and Eosin-Stained Whole Slide Images

TL;DR

This paper presents a federated learning approach combined with a point transformer model to accurately predict HER2 status from WSIs, addressing data imbalance and generalization across multiple sites.

Contribution

It introduces a novel point transformer architecture with dynamic label handling and sampling strategies for federated learning in medical imaging.

Findings

Achieved state-of-the-art results on four sites with 2687 WSIs.

Demonstrated strong generalization to two unseen sites with 229 WSIs.

Improved prediction accuracy and robustness in multi-site datasets.

Abstract

This study introduces a federated learning-based approach to predict HER2 status from hematoxylin and eosin (HE)-stained whole slide images (WSIs), reducing costs and speeding up treatment decisions. To address label imbalance and feature representation challenges in multisite datasets, a point transformer is proposed, incorporating dynamic label distribution, an auxiliary classifier, and farthest cosine sampling. Extensive experiments demonstrate state-of-the-art performance across four sites (2687 WSIs) and strong generalization to two unseen sites (229 WSIs).