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

Conference · Fri Jan 01 00:00:00 EST 2010

OSTI ID:1004432

Kerekes, Ryan A ^[1]; Gleason, Shaun Scott ^[1]; Martins, Rodrigo ^[2]; Dyer, Michael ^[2]

---

1. ORNL
2. St. Jude Children's Research Hospital

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.

• Research Organization: Oak Ridge National Lab. (ORNL), Oak Ridge, TN (United States)
• Sponsoring Organization: USDOE Laboratory Directed Research and Development (LDRD) Program
• DOE Contract Number: DE-AC05-00OR22725
• OSTI ID: 1004432
• Resource Relation: Conference: Biomedical Science and Engineering Conference (BSEC 2010), Oak Ridge, TN, USA, 20100525, 20100526
• Country of Publication: United States
• Language: English

Related Subjects

59 BASIC BIOLOGICAL SCIENCES
99 GENERAL AND MISCELLANEOUS//MATHEMATICS, COMPUTING, AND INFORMATION SCIENCE
ALGORITHMS
NERVE CELLS
RETINA
IMAGES
PERFORMANCE