A transfer learning approach to few-shot segmentation of novel white matter tracts
Medical Image Analysis(2022)
摘要
•We explore knowledge transfer for few-shot segmentation of novel WM tracts.•A model pretrained for segmenting existing WM tracts is given.•We derive a better network initialization for segmenting novel WM tracts.•Data augmentation is further applied with tract-aware image mixing.•Our method was validated on public and private datasets under various settings.
更多查看译文
关键词
White matter tract segmentation,Novel white matter tract,Few-shot segmentation,Transfer learning
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要