Scholarlink

We introduce Mysh, an automated tool for instance segmentation and morphometric analysis of peripheral nerve light microscopy images. It employs a two-stage pipeline: a U-Net first segments axons, followed by separation using Connected Component Labeling. Then, a second U-Net segments the myelin for each individual axon. While computationally demanding, Mysh outperforms AxonDeepSeg on all semantic segmentation metrics and enables precise morphometric analysis at the single-axon level.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Automated Instance Segmentation of Peripheral Nerve Microscopy Images: A Novel Pipeline for Morphometric Analysis in Regenerative Medicine

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Krzysztof Spaliński,
  • Antonina Spalińska,
  • Michal Kopka,
  • Paulina Przybylek,
  • Magdalena Jasionowska-Skop,
  • Artur Przelaskowski

摘要

We introduce Mysh, an automated tool for instance segmentation and morphometric analysis of peripheral nerve light microscopy images. It employs a two-stage pipeline: a U-Net first segments axons, followed by separation using Connected Component Labeling. Then, a second U-Net segments the myelin for each individual axon. While computationally demanding, Mysh outperforms AxonDeepSeg on all semantic segmentation metrics and enables precise morphometric analysis at the single-axon level.