BUILDING A NEURON ORIENTED DATABASE OF THE RETINA

描述（改编自申请人摘要）：我们建议继续我们的

DESCRIPTION (Adapted from applicant's abstract): We propose to continue our efforts towards the design and implementation of a database of the retina. To date, our work on database schema design and implementation has made us acutely aware of the considerable complexity and importance in the systematic classification and organization of cellular neuroscience data. Prior to developing a comprehensive neurosciences database a variety of issues in neuron classification using morphological and physiological properties must be addressed. This proposal considers these issues via the development of a suite of software tools to support improved data collection and visualization methodology, morphological and feature extraction from confocal images and the enumeration of a set of physiological features for inclusion in a database. Specifically, we propose a research effort that provides the following items: (1) Data Collection and Visualization - An improved method for collection of morphological and physiological data from intracellularly filled cells using the confocal microscope is proposed. We will also extend our MacVol volume visualization program to handle double and triple labeled datasets produced by the confocal microscope; (2)Classification (Morphology)- We will continue our efforts on the identification and extraction of morphological features from neuronal images. Empirical investigation of the use of these image features to support content-based retrieval operations in a database of such images will be conducted; (3) Classification (Physiology)-Our current database schema will be extended to more adequately handle the complexities of physiological data. Of specific interest are the kinds of data extracted using the methodology developed in the context of item (1) above; and (4) Data Comparison via a Neuron Browser-A library of neuronal images along with physiological descriptors will be developed. Software tools for querying and browsing images will also be implemented and made available.