MedFL-Stress: A Systematic Robustness Evaluation of Federated Brain Tumor Segmentation under Cross-Hospital MRI Appearance Shift

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.

Purpose: Developing generalizable medical image segmentation models is challenging because imaging data are distributed across institutions and differ in modality and acquisition protocol. Federated learning (FL) enables collaborative training without centralizing raw medical images, but cross-modality domain shifts between computed tomography (CT) and magnetic resonance imaging (MRI) can substantially reduce model performance. This study investigates augmentation-driven cross-modality FL for abdominal organ and whole-heart segmentation. Methods: We evaluate convolution-based spatial augmentation, frequency-domain argumentation, domain-specific normalization, and global intensity nonlinear (GIN) augmentation for multimodal segmentation. Abdominal organ segmentation and whole-heart segmentation are first evaluated using a 2D U-Net framework. For whole-heart segmentation, we additionally perform native 3D experiments using a self-configuring nnU-Net architecture on the CARE-WHS 2026 dataset, enabling evaluation of whether the observed cross-modality FL behavior persists when moving from slice-based 2D segmentation to volumetric 3D segmentation. Results: GIN provides the most consistent cross-modality performance among the evaluated approaches in the original 2D experiments. For pancreas segmentation, the Dice similarity coefficient (DSC) improved from 0.073 to 0.437 when CT data were incorporated through federated cross-modality training. In 3D whole-heart segmentation, FedGIN improved mean DSC over FedAvg from 0.8696 to 0.8901 on the unseen CT center and from 0.7160 to 0.7956 on the unseen MRI center. Relative to centralized GIN training, FedGIN retained 92.4% of performance on unseen CT data and achieved comparable performance on unseen MRI data (0.7956 versus 0.7937).

Federated Learning (FL) enables collaborative training of machine learning models across multiple institutions without sharing sensitive data, making it particularly suitable for medical imaging applications. However, heterogeneous data distributions across institutions and potential information leakage through model updates remain important challenges. In this work, we propose DP-SimAgg, a privacy-preserving federated learning framework that integrates similarity-weighted aggregation with a server-side differential privacy mechanism. The proposed method applies L2 clipping to bound collaborator updates, computes similarity-based aggregation weights to mitigate the effects of non-IID data distributions, and injects calibrated Gaussian noise at the central server, providing per-round privacy guarantees under the assumed sensitivity bound. The framework is implemented using Intel's OpenFL platform and evaluated on the FeTS 2022 dataset consisting of 1251 multi-modal MRI scans for brain tumor segmentation. Experimental results demonstrate that DP-SimAgg maintains competitive segmentation performance while providing privacy protection. Under a strict per-round privacy budget (epsilon = 1, cumulative epsilon_total = 20 over 20 rounds), the method achieves Dice scores of 0.6357, 0.5305, and 0.5274 for the enhancing tumor (ET), tumor core (TC), and whole tumor (WT) regions, respectively. With a more relaxed per-round budget (epsilon = 10, cumulative epsilon_total = 200), performance approaches that of the non-private baseline while incorporating a central Gaussian mechanism with per-round (epsilon, delta)-DP accounting under the assumed sensitivity bound. These results highlight the potential of DP-SimAgg for enabling privacy-preserving collaborative learning in medical imaging applications.