An exciting and emerging area of biological research is centered on the creation of neuronal atlases, often referred to as neuromes, which provide structural description of all neurons in a particular organism. Aside from tedious manual analysis, there exist no reliable methods to extract structural knowledge regarding a neuron from a 3D image. This project will develop fully automated informatics tools that will allow the biologist to automatically extract 3D structure from the images. The Tree2Tree algorithm converts 3D images of neuronal trees to graph theoretic trees that can be represented efficiently by computers and compared to enable retrieval and dissemination of structural information regarding an organism's neurome. Tree2Tree features novel techniques for extracting the tubular structure of the neurons, for performing connected component analysis of the neurite portions, for representing the neuron as a graph theoretic tree, for pruning the tree and for comparing the resultant trees in a database of neurons.
A cross-disciplinary team from Biology and Engineering has been assembled to tackle the research problem of creating a neuronal atlas. The innovations in software will open doors to creating new atlases for complex organisms, which can be in turn used in important developmental and behavioral studies. These informatics tools will serve as a platform from which advances in biological discovery can occur. All proposed informatics tools will be disseminated as open source code and plug-ins in three widely used repositories as well as at http://viva.ee.virginia.edu/research.html. The final product of this research will include informatics tools as well as a working database example using the fruit fly.
