Yipeng Hu

Continual pretraining a ViT encoder with LoRA on prostate MRI segmentation significantly improves downstream task alignment and performance, particularly for segmentation tasks, by inducing targeted representational shifts.

Actively repelling uncertain features during representation learning slashes extreme segmentation errors in ambiguous anatomical boundaries by 22x, even without sacrificing overall accuracy.

By adapting model contributions to specific anatomical targets and institutional contexts, this framework significantly enhances segmentation performance in environments with scarce labeled data.