Federated Stain Normalization for Computational Pathology

TL;DR

This paper introduces BottleGAN, a generative model designed to align staining styles across laboratories in computational pathology, enabling privacy-preserving federated learning and significantly improving segmentation accuracy.

Contribution

The paper presents BottleGAN, a novel style alignment model for federated learning in pathology, addressing the challenge of heterogeneous staining styles across labs.

Findings

Improved IOU by 42% over existing federated algorithms.

Constructed a multi-institutional dataset based on PESO.

Demonstrated effective style alignment in computational pathology.

Abstract

Although deep federated learning has received much attention in recent years, progress has been made mainly in the context of natural images and barely for computational pathology. However, deep federated learning is an opportunity to create datasets that reflect the data diversity of many laboratories. Further, the effort of dataset construction can be divided among many. Unfortunately, existing algorithms cannot be easily applied to computational pathology since previous work presupposes that data distributions of laboratories must be similar. This is an unlikely assumption, mainly since different laboratories have different staining styles. As a solution, we propose BottleGAN, a generative model that can computationally align the staining styles of many laboratories and can be trained in a privacy-preserving manner to foster federated learning in computational pathology. We construct a heterogenic multi-institutional dataset based on the PESO segmentation dataset and improve the IOU by 42\% compared to existing federated learning algorithms. An implementation of BottleGAN is available at https://github.com/MECLabTUDA/BottleGAN