CEPerFed: Communication-Efficient Personalized Federated Learning for Multi-Pulse MRI Classification

2025-10-20

AI Summary

To address strong data heterogeneity, high communication overhead, and urgent privacy requirements in multi-center, multi-pulse MRI classification, this paper proposes CEPerFed—a communication-efficient personalized federated learning framework. Methodologically: (1) it introduces a weighted aggregation of client-specific historical risk gradients and mean gradients to enhance local update robustness while preserving global model consistency; (2) it designs a hierarchical SVD-based parameter compression strategy to drastically reduce uplink bandwidth consumption. Evaluated on five cross-institutional MRI classification tasks, CEPerFed achieves an average 2.3% accuracy gain over baseline methods, accelerates convergence by 37%, reduces total communication volume by 68%, and satisfies rigorous differential privacy guarantees. This work delivers a practical, privacy-preserving solution for distributed medical AI that jointly optimizes performance, communication efficiency, and regulatory compliance.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Computer Vision: Multi-modal Vision
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Security and Privacy: Security and privacy of machine learning and AI applications
• Search and Retrieval-Augmented AI: Personalized, context-aware and across-device search

Abstract

Multi-pulse magnetic resonance imaging (MRI) is widely utilized for clinical practice such as Alzheimer's disease diagnosis. To train a robust model for multi-pulse MRI classification, it requires large and diverse data from various medical institutions while protecting privacy by preventing raw data sharing across institutions. Although federated learning (FL) is a feasible solution to address this issue, it poses challenges of model convergence due to the effect of data heterogeneity and substantial communication overhead due to large numbers of parameters transmitted within the model. To address these challenges, we propose CEPerFed, a communication-efficient personalized FL method. It mitigates the effect of data heterogeneity by incorporating client-side historical risk gradients and historical mean gradients to coordinate local and global optimization. The former is used to weight the contributions from other clients, enhancing the reliability of local updates, while the latter enforces consistency between local updates and the global optimization direction to ensure stable convergence across heterogeneous data distributions. To address the high communication overhead, we propose a hierarchical SVD (HSVD) strategy that transmits only the most critical information required for model updates. Experiments on five classification tasks demonstrate the effectiveness of the CEPerFed method. The code will be released upon acceptance at https://github.com/LD0416/CEPerFed.

Problem

• Addressing data heterogeneity in federated MRI classification
• Reducing communication overhead in distributed medical learning
• Ensuring stable convergence across heterogeneous client data

Innovation

• Uses client-side historical gradients for local optimization
• Enforces consistency between local updates and global direction
• Employs hierarchical SVD to reduce communication overhead