Federated Learning for Medical Image Analysis: A Survey

TL;DR

This survey addresses the privacy-driven, small-sample challenge in medical image analysis by examining federated learning as a collaborative training paradigm. It presents a system-oriented taxonomy that splits methods into client-end, server-end, and communication solutions, and highlights practical platforms and datasets that enable FL research. Through an experimental comparison on the ADNI dataset, the paper shows that federated strategies generally outperform isolated training, with mixed data approaches offering substantial gains and weight-averaging methods often yielding robust performance. The work identifies key challenges such as domain shift, data heterogeneity, and security threats while outlining concrete future directions, including multi-modality FL, personalized strategies, and larger-scale benchmarks to accelerate clinical impact.

Abstract

Machine learning in medical imaging often faces a fundamental dilemma, namely, the small sample size problem. Many recent studies suggest using multi-domain data pooled from different acquisition sites/centers to improve statistical power. However, medical images from different sites cannot be easily shared to build large datasets for model training due to privacy protection reasons. As a promising solution, federated learning, which enables collaborative training of machine learning models based on data from different sites without cross-site data sharing, has attracted considerable attention recently. In this paper, we conduct a comprehensive survey of the recent development of federated learning methods in medical image analysis. In this survey, we first introduce the background knowledge of federated learning for dealing with privacy protection and collaborative learning issues in medical imaging. We then present a comprehensive review of recent advances in federated learning methods for medical image analysis. Specifically, existing methods are categorized based on three critical aspects of a federated learning system, including client end, server end, and communication techniques. In each category, we summarize the existing federated learning methods according to specific research problems in medical image analysis and also provide insights into the motivations of different approaches. In addition, we provide a review of existing benchmark medical imaging datasets and software platforms for current federated learning research. We also conduct an experimental study to empirically evaluate typical federated learning methods for medical image analysis. This survey can help to better understand the current research status, challenges, and potential research opportunities in this promising research field.

Figures (7)

• Figure 1: Overview of federated learning (FL) for medical image analysis, including a server and multiple clients. Each selected client trains a model on its local dataset. The server collects the local models and calculates a global model that is broadcast to all the selected clients for deployment.
• Figure 2: Overview of the number of papers (in terms of published years) that have been collected for this survey on federated learning in medical image analysis.
• Figure 3: Overview of federated learning (FL) methods for medical image analysis.
• Figure 4: Domain shift among different medical sites (domains). Domain adaptation aims to reduce domain differences and enhance machine learning performance across different sites. Image courtesy to Guan et al.guan2023domainatm.
• Figure 5: Different local updates for clients with different computation and data resources.