Improving segmentation of 3D touching cell nuclei using flow tracking on surface meshes

Int J Comput Biol Drug Des. 2012;5(1):66-74. doi: 10.1504/IJCBDD.2012.045952. Epub 2012 Mar 21.

Abstract

Automatic segmentation of touching cell nuclei in 3D microscopy images is of great importance in bioimage informatics and computational biology. This paper presents a novel method for improving 3D touching cell nuclei segmentation. Given binary touching nuclei by the method in Li et al. (2007), our method herein consists of several steps: surface mesh reconstruction and curvature information estimation; direction field diffusion on surface meshes; flow tracking on surface meshes; and projection of surface mesh segmentation to volumetric images. The method is validated on both synthesised and real 3D touching cell nuclei images, demonstrating its validity and effectiveness.

MeSH terms

  • Animals
  • Cell Nucleus / ultrastructure*
  • Diffusion
  • Imaging, Three-Dimensional / methods*
  • Zebrafish