Automated Morphometry of Dendritic Spines

Suny At Stony Brook, Stony Brook NY

Investigators

Abstract

Lindquist 0107893 The investigator creates quantitative computational tools for the automated analysis of neural images, including software algorithms to provide automated morphology measurements both of neuron dendrites and of dendritic spines (the primary synaptic receivers). These tools are developed in close collaboration with the neural imaging research programs of two groups, led by Prof. K. Svoboda at Cold Spring Harbor Lab. and by Profs. P. Hof and S. Wearne at Mt. Sinai School of Medicine. In preliminary work, automated algorithms have been developed and verified (against manual measurements) to identify and extract morphological measurements of dendritic spine length and density from 3D images. This project develops algorithms to i) provide spine volume measurements, which are currently available only through manually intensive, scanning electron microscopy measurements; ii) analyze (two-channel) images containing two dyes that fluoresce at different frequencies. One dye is a marker for spine morphology, the second is a marker for specific functionality. This allows for direct correlation studies of form and function. The primary application of these algorithms is in structural genetics work. iii) determine higher moment (shape) measures both of spines and dendrite cross sections. The primary application of these algorithms is in developing models of neural integration behavior and age-related deficits in short-term memory. Biological science technologies are now capable of producing data at a rate so fast that only with computers and sophisticated numerical and statistical algorithms can biologists analyze these data. In the past decade, laser scanning fluorescence microscopy, enabling non-invasive, three dimensional imaging of living neurons, has produced a revolution in neural research. The wealth of information available from a time sequence of three dimensional neuron images is outstripping our current information processing capacity, which has been based upon manual identification and tracing of structures of interest. Although it uses the superb recognition capability of human sight, manual measurement is tedious, susceptible to systematic and hard to characterize human biases, restricted to measurements involving only length and counting, and effective only for small data sets. The purpose of this project is to develop computer based tools for the automated analysis of neuron images, eliminating manual analysis. The resultant computer tools will make practical such studies as i) the relationship between neuron structure and age-related deficits in memory as well as some forms of mental retardation; ii) where in neurons certain proteins concentrate and how their presence and concentration correlates to the signal processing role of neurons; iii) large scale screening of drug treatments designed to correct genetic- or trauma-induced deficiencies in neuron structure. The project is supported by the DMS/Computational Mathematics, DMS/Applied Mathematics, and IBN/Computational Neuroscience programs and by the MPS/Office of Multidisciplinary Activities.