Federated Medical Image Classification under Class and Domain Imbalance exploiting Synthetic Sample Generation
For medical institutions collaborating under privacy constraints, this method mitigates the combined challenges of domain shift and class imbalance without requiring additional local data.
FedSSG addresses class and domain imbalance in federated medical image classification by generating and distributing synthetic samples across clients, significantly improving model performance and generalization with minimal computational overhead.
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.