From Gradient Clipping to Structural Refinement: Improving DPSGD for Medical Image Segmentation

2026-06-19

AI Summary

This work addresses the significant performance degradation in differentially private (DP) medical image segmentation by systematically evaluating various DP-SGD variants and uncovering the limitations of existing adaptive gradient clipping methods in multi-class segmentation tasks. To mitigate this issue, the authors propose DP-Morph, a class-aware approach that integrates gradient alignment analysis with morphological structure optimization to enhance segmentation accuracy under stringent privacy constraints. Experimental results demonstrate that DP-Morph consistently outperforms current state-of-the-art methods in both binary and multi-class settings, achieving notably superior performance in regions with complex anatomical structures. The proposed method thus effectively alleviates the trade-off between privacy preservation and model utility in medical image segmentation.

Abstract

Medical image segmentation is widely used for disease detection but relies on sensitive data, raising privacy concerns as trained models can leak information. Differential privacy, typically implemented via Differential Private Stochastic Gradient Descent (DPSGD), provides a solution, though at the cost of reduced utility. Recent DPSGD variants, including Automatic clipping (Auto-S), Normalised SGD with perturbation (NSGD), and Per-sample adaptive clipping (PSAC), have shown promise in image classification, but their behavior in medical segmentation remains underexplored. We evaluate these methods across binary and multi-class tasks and analyze gradient alignment, showing that prior assumptions, particularly for PSAC, do not consistently hold. We further demonstrate that combining clipping strategies with morphological refinement improves segmentation quality under privacy constraints. Finally, we propose an adaptive DP-Morph variant that captures class-specific structures and enhances performance in multi-class settings.

Problem

• differential privacy
• medical image segmentation
• DPSGD
• utility degradation
• privacy-preserving

Innovation

• Differential Privacy
• Medical Image Segmentation
• Gradient Clipping
• Morphological Refinement
• Adaptive DP-Morph