Medical Image Segmentation: Foundation Model Adaptation versus Task-Specific Training

arXiv:2611.14519v1 Announce Type: cross Abstract: Foundation models pretrained on large natural-image collections are increasingly adapted for medical image segmentation, in the expectation that they will replace models trained for each task.