Automated 3d Reconstruction of Neuronal Structures from Serial Sections

Automated 3D Reconstruction of Neuronal Structures from Serial Sections. (August 1999) Brent P. Burton, B.S., Texas A&M University Chair of Advisory Committee: Dr. Bruce H. McCormick A fast automated system for tracing neurons in parallel is described, adequate to support a quantitative analysis of neuron morphology. The system described here automates digitized neuron feature extraction and reconstruction, thereby replacing current largely manual techniques for tracing individual neurons. Serial sections of brain tissue are created by physical sectioning. Sections are processed during scanning to determine regions of interest (ROIs) and to quickly cull unnecessary image data. An aggressive data culling and compression scheme reduces the original volumetric data into a ROI-based image collection that makes temporary secondary storage feasible. Neighboring ROIs are then matched from successive sections for segment and ber tracing. Reconstructed segments created from these matches are disambiguated, resulting in an abbreviated structural description of the tissue's neurons and ber tracts.