Adaptive Differential Privacy for Federated Medical Image Segmentation Across Diverse Modalities

2026-04-07

AI Summary

This work addresses the challenges of degraded accuracy, unstable convergence, and poor generalization in federated medical image segmentation caused by data heterogeneity and stringent privacy constraints. To tackle these issues, the authors propose an Adaptive Differential Privacy Federated Learning framework (ADP-FL), which dynamically adjusts the noise intensity of differential privacy based on training dynamics while preserving raw data locality. Integrating a U-Net variant architecture, ADP-FL achieves an optimal trade-off between privacy guarantees and model utility. Extensive experiments on multimodal 2D/3D medical image segmentation tasks—including skin lesions, kidney tumors, and brain tumors—demonstrate that ADP-FL significantly outperforms conventional federated learning and standard differentially private approaches in terms of Dice score, boundary quality, convergence speed, and training stability, achieving performance close to that of non-private federated baselines.

Abstract

Large volumes of medical data remain underutilized because centralizing distributed data is often infeasible due to strict privacy regulations and institutional constraints. In addition, models trained in centralized settings frequently fail to generalize across clinical sites because of heterogeneity in imaging protocols and continuously evolving data distributions arising from differences in scanners, acquisition parameters, and patient populations. Federated learning offers a promising solution by enabling collaborative model training without sharing raw data. However, incorporating differential privacy into federated learning, while essential for privacy guarantees, often leads to degraded accuracy, unstable convergence, and reduced generalization. In this work, we propose an adaptive differentially private federated learning (ADP-FL) framework for medical image segmentation that dynamically adjusts privacy mechanisms to better balance the privacy-utility trade-off. The proposed approach stabilizes training, significantly improves Dice scores and segmentation boundary quality, and maintains rigorous privacy guarantees. We evaluated ADP-FL across diverse imaging modalities and segmentation tasks, including skin lesion segmentation in dermoscopic images, kidney tumor segmentation in 3D CT scans, and brain tumor segmentation in multi-parametric MRI. Compared with conventional federated learning and standard differentially private federated learning, ADP-FL consistently achieves higher accuracy, improved boundary delineation, faster convergence, and greater training stability, with performance approaching that of non-private federated learning under the same privacy budgets. These results demonstrate the practical viability of ADP-FL for high-performance, privacy-preserving medical image segmentation in real-world federated settings.

Problem

• Federated Learning
• Differential Privacy
• Medical Image Segmentation
• Privacy-Utility Trade-off
• Multi-modality

Innovation

• adaptive differential privacy
• federated learning
• medical image segmentation
• privacy-utility trade-off
• heterogeneous modalities