Transfer Learning Approaches for Medical Image Segmentation in Low-Data Regimes

Abstract

Medical image segmentation in resource-limited settings requires models that perform well with limited labeled data. We propose a two-stage transfer learning strategy.