Uncertainty-Informed Collaborative Learning (UnCoL)

UnCoL is a dual-teacher framework for semi-supervised medical image segmentation. It harmonizes general knowledge from foundation models with task-specific representations to improve segmentation performance under limited annotations.