FedOnco-Bench: A Reproducible Benchmark for Privacy-Aware Federated Tumor Segmentation with Synthetic CT Data
This work addresses privacy and data heterogeneity issues in federated learning for medical applications, providing a standardized benchmark for researchers and developers, but it is incremental as it builds on existing FL methods.
The paper tackles the problem of privacy leakage and data heterogeneity in federated learning for medical image segmentation by introducing FedOnco-Bench, a reproducible benchmark using synthetic CT data, showing a trade-off where FedAvg achieves high segmentation performance (Dice around 0.85) but more privacy leakage (attack AUC about 0.72), while DP-SGD provides better privacy (AUC around 0.25) at reduced accuracy (Dice about 0.79).
Federated Learning (FL) allows multiple institutions to cooperatively train machine learning models while retaining sensitive data at the source, which has great utility in privacy-sensitive environments. However, FL systems remain vulnerable to membership-inference attacks and data heterogeneity. This paper presents FedOnco-Bench, a reproducible benchmark for privacy-aware FL using synthetic oncologic CT scans with tumor annotations. It evaluates segmentation performance and privacy leakage across FL methods: FedAvg, FedProx, FedBN, and FedAvg with DP-SGD. Results show a distinct trade-off between privacy and utility: FedAvg is high performance (Dice around 0.85) with more privacy leakage (attack AUC about 0.72), while DP-SGD provides a higher level of privacy (AUC around 0.25) at the cost of accuracy (Dice about 0.79). FedProx and FedBN offer balanced performance under heterogeneous data, especially with non-identical distributed client data. FedOnco-Bench serves as a standardized, open-source platform for benchmarking and developing privacy-preserving FL methods for medical image segmentation.