FedMedICL: Towards Holistic Evaluation of Distribution Shifts in Federated Medical Imaging

TL;DR

FedMedICL introduces a comprehensive benchmark for evaluating federated medical imaging models under multiple simultaneous distribution shifts, revealing that simple techniques can outperform complex methods in real-world scenarios.

Contribution

This work presents FedMedICL, the first unified framework to evaluate multiple co-occurring distribution shifts in federated medical imaging, highlighting the limitations of current benchmarks.

Findings

Simple batch balancing outperforms advanced methods

Current benchmarks may overestimate model robustness

Simulating COVID-19 spread reveals adaptation challenges

Abstract

For medical imaging AI models to be clinically impactful, they must generalize. However, this goal is hindered by (i) diverse types of distribution shifts, such as temporal, demographic, and label shifts, and (ii) limited diversity in datasets that are siloed within single medical institutions. While these limitations have spurred interest in federated learning, current evaluation benchmarks fail to evaluate different shifts simultaneously. However, in real healthcare settings, multiple types of shifts co-exist, yet their impact on medical imaging performance remains unstudied. In response, we introduce FedMedICL, a unified framework and benchmark to holistically evaluate federated medical imaging challenges, simultaneously capturing label, demographic, and temporal distribution shifts. We comprehensively evaluate several popular methods on six diverse medical imaging datasets (totaling 550 GPU hours). Furthermore, we use FedMedICL to simulate COVID-19 propagation across hospitals and evaluate whether methods can adapt to pandemic changes in disease prevalence. We find that a simple batch balancing technique surpasses advanced methods in average performance across FedMedICL experiments. This finding questions the applicability of results from previous, narrow benchmarks in real-world medical settings.