Federated Medical Image Classification under Class and Domain Imbalance exploiting Synthetic Sample Generation

2026-04-29

AI Summary

This work addresses the challenges of domain shift caused by device heterogeneity and the scarcity of rare disease samples in medical image federated learning. To tackle these issues, the authors propose FedSSG, a framework that enables clients to generate and share lightweight synthetic samples, thereby collaboratively mitigating both class imbalance and cross-domain distribution discrepancies while preserving data privacy. By seamlessly integrating synthetic data augmentation into the federated training pipeline, FedSSG significantly enhances model classification performance and generalization across multi-center heterogeneous settings with minimal computational overhead, particularly improving recognition accuracy for rare disease classes.

Abstract

Exploiting deep learning in medical imaging faces critical challenges, including strict privacy constraints, heterogeneous imaging devices with varying acquisition properties, and class imbalance due to the uneven prevalence of pathologies. In this work, we propose FedSSG, a novel Federated Learning framework that addresses domain shifts caused by diverse imaging devices while mitigating the under-representation of rare pathologies. The key contribution is a strategy for generating synthetic samples and distributing them across clients to improve coverage of both underrepresented pathologies and imaging devices. Experimental results demonstrate that our approach significantly enhances model performance and generalization across heterogeneous institutions, with minimal computational overhead at the client side.

Problem

• federated learning
• medical image classification
• class imbalance
• domain shift
• synthetic sample generation

Innovation

• Federated Learning
• Synthetic Sample Generation
• Domain Imbalance
• Class Imbalance
• Medical Image Classification