Federated Medical Image Segmentation under Real-World Label Noise: A Benchmark Suite for Noisy Label Learning Method Selection

2026-06-15

AI Summary

This work addresses the performance degradation in federated medical image segmentation caused by real-world annotation noise—such as inconsistent contours, missing anatomical structures, or label ambiguities—by introducing the first benchmark suite specifically designed for realistic noise scenarios. The proposed suite integrates multi-source medical images with authentic annotation imperfections, client-wise noise configurations that mirror practical deployment settings, and dedicated evaluation metrics, thereby establishing a reproducible, fair, and discriminative assessment framework within federated learning. By providing a realistic, reliable, and scalable foundation for developing and evaluating noise-robust methods, this study significantly enhances the scientific rigor of algorithm selection and supports community advancement through open-sourced code.

Abstract

While federated learning (FL) enables collaborative medical image segmentation without centralizing sensitive data, real-world deployment is frequently complicated by cross-site label imperfections such as contour disagreement, missing or additional structures, and confused labels. Federated noisy label learning (FNLL) aims to mitigate these effects, yet remains underused in practice as existing evidence is largely based on synthetic noise, simplified settings, and limited real-world noisy evaluation. We address this gap by introducing a benchmark suite that combines diverse real-world noisy datasets, deployment-relevant client-noise scenarios, and label-noise-targeted evaluation to support systematic FNLL assessment and informed method selection. The suite combines curated real-world noisy medical image segmentation datasets from diverse sources with a comprehensive federated segmentation framework including various client-noise scenarios and noise-targeted evaluation. The presented suite provides a realistic and discriminative basis for FNLL evaluation in medical image segmentation and establishes a reusable foundation for fair benchmarking, dataset-specific label-noise characterization, and future method development under realistic federated settings. Code is available at https://github.com/MIC-DKFZ/FedSegNoiseBench.

Problem

• federated learning
• medical image segmentation
• label noise
• noisy label learning
• real-world noise

Innovation

• federated learning
• medical image segmentation
• label noise
• benchmark suite
• noisy label learning