Federated Tumor Segmentation (FeTS) Challenge

The FeTS 2021 challenge focuses on glioma segmentation under federated learning, with two tasks: finding the best weight aggregation method to train a consensus model, and cross-institution generalization evaluation. It uses clinical multi-institutional MRI scans from BraTS 2020 and data from remote independent institutions in a real federated collaboration network, with data always kept at each institution during training. Generalization evaluation is performed on data distributions from institutions that did not participate in training, examining the performance of brain tumor segmentation models in real-world settings.