Scholarlink

Somatic mutations are known to be the root cause of cancer development and progression. However, detecting mutations in single cells or clones remains a challenging task due to the high error rates and restrictions on specific cell types. With the recent development of such protocol as NanoSeq it is now possible to reliably detect mutations in single DNA molecules, which could transform understanding of somatic mutagenesis and enable non-invasive studies on large-scale cohorts. Thus, we are in urgent need of large-scale studies of human cohorts of different ages, occupations, lifestyles, and health statuses. One of the promising non-invasive approaches is collecting buccal swab samples to study epithelial cell mutations. However, it is complicated by the need to establish a robust quality control pipeline of microscopy image data for cell count estimation in a given sample to avoid possible contamination. Such pipeline should be able to process hundreds of images in a reasonable time and accurately detect cells of a certain type automatically. We present the Automated Buccal Swabs Cells Recognition (ABSCR) tool that allows for the automated segmentation of human epithelial cells from buccal swabs microscopy image data using GPU-enabled Deep Learning methods and integration with the OMERO platform, which is widely used for managing, visualising and analysing microscopy images and associated metadata.

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

Pipeline for Automatic Segmentation of Epithelial Cells in Buccal Swabs Slides with Custom Cellpose-Based Model

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Dmytro Horyslavets,
  • Oleksandr Skorobohatov,
  • Yaroslav Ryndyk,
  • Pavlo Areshkov,
  • Olena Romantsova,
  • Charlotte Oliver,
  • Anna Paterson,
  • Wei Cope,
  • Emily Joslin,
  • Emily O’Dea,
  • Jon Teague,
  • Peter Clapham,
  • Moritz Przybilla,
  • Mykhailo Tukalo,
  • James McCafferty,
  • Inigo Martincorena,
  • Alina Frolova

摘要

Somatic mutations are known to be the root cause of cancer development and progression. However, detecting mutations in single cells or clones remains a challenging task due to the high error rates and restrictions on specific cell types. With the recent development of such protocol as NanoSeq it is now possible to reliably detect mutations in single DNA molecules, which could transform understanding of somatic mutagenesis and enable non-invasive studies on large-scale cohorts. Thus, we are in urgent need of large-scale studies of human cohorts of different ages, occupations, lifestyles, and health statuses. One of the promising non-invasive approaches is collecting buccal swab samples to study epithelial cell mutations. However, it is complicated by the need to establish a robust quality control pipeline of microscopy image data for cell count estimation in a given sample to avoid possible contamination. Such pipeline should be able to process hundreds of images in a reasonable time and accurately detect cells of a certain type automatically. We present the Automated Buccal Swabs Cells Recognition (ABSCR) tool that allows for the automated segmentation of human epithelial cells from buccal swabs microscopy image data using GPU-enabled Deep Learning methods and integration with the OMERO platform, which is widely used for managing, visualising and analysing microscopy images and associated metadata.