VerSemi

VerSemi is a framework for semi-supervised medical image segmentation that integrates various SSL tasks into a unified model, exploiting more unlabeled data. It leverages a dynamic task-prompted design to segment various targets from different datasets, consistently outperforming existing methods.