Skip to main content

Fully automated segmentation and characterization of the dendritic trees of retinal horizontal neurons...

by Ryan A Kerekes, Shaun S Gleason, Rodrigo Martins, Michael Dyer
Publication Type
Conference Paper
Publication Date
Conference Name
Biomedical Science and Engineering Conference (BSEC 2010)
Conference Location
Oak Ridge, Tennessee, United States of America
Conference Sponsor
Conference Date

We introduce a new fully automated method for segmenting and characterizing the dendritic tree of neurons in confocal image stacks. Our method is aimed at wide-field-of-view, low-resolution imagery of retinal neurons in which dendrites can be intertwined and difficult to follow. The approach is based on 3-D skeletonization and includes a method for automatically determining an appropriate global threshold as well as a soma detection algorithm. We provide the details of the algorithm and a qualitative performance comparison against a commercially available neurite tracing software package, showing that a segmentation produced by our method more closely matches the ground-truth segmentation.