University of Houston
Department of Computer Science

 In partial fulfillment of the Requirements for the Degree of
 Master of Science
 
 
 Steve Urban
will defend his thesis
 
 Automatic Reconstruction of Dendrite Morphologies from  Optical Sections of Living Fluorescently Labeled Neurons

Abstract
 The function of the human brain arises from computations that occur within  and among millions of nerve cells. Each nerve cell receives information at  its synapses, which are usually situated on dendritic spines. This information  is processed by the dendrites' specific morphology and distribution of voltage-gated  ion channels. To understand these dendritic computations, it is necessary  to analyze the complex interaction between structure and function. An important  step in this process is to acquire an accurate morphological reconstruction  of a cell. Currently, researchers rely on manual or semi-automatic methods  to reconstruct morphologies from brightfield or fluorescent images of neurons.  Manual methods are usually very tedious and subject to user bias. Most importantly,  they are so time consuming that the reconstruction of live neurons is not  possible. While some semi-automatic methods exist, they still require a high  level of user interaction. In collaboration with the UH Department of Biology  and Biochemistry and Baylor College of Medicine, the Computational Biomedicine  Lab has developed a new automated neuron reconstruction system, NeuRecon,  that extracts the morphological structure of neuronal data from confocal and  multi-photon microscopes with minimal human interaction. This system employs  several novel techniques that for the first time allow for the automated morphological  reconstruction of dendrites. We present an overview of the NeuRecon system,  with emphasis on the registration and spine segmentation methods. The results  of this work will guide the efficient design of experiments and the generation  of hypotheses to reveal functional mechanisms underlying both normal and diseased  states of the nervous system.

 Date: Monday, August 8, 2005
 Time: 11:00 AM
 Place: 218D-PGH
 
 
 
 Faculty, students, and the general public are invited.
 Thesis Advisor: Dr. Kakadiaris