Recent advances in understanding neuronal functions have led to an expanded scientific interest in defining the organization of the nervous system so the spatial correlates of these functions can be identified. This interest has been formalized as the Human Brain Project, an ambitious multidisciplinary effort to map the nervous system from the organismal to the macromolecular levels. One of the greatest challenges of this effort is to preserve the complex three-dimensional relationships that occur between neuronal structures. This problem will require new methods for data acquisition as well as data viscualization. The project described here is an interdisciplinary effort to derive three-dimensional reconstructions of synaptic architecture form stereo electron micrographs acquired from multiple viewpoints. The collaboration combines advanced ultrastructural visualization techniques with massively parallel computational methods and an innovative set of pattern recognition, stereo correspondence and depth mapping algorithms. Our goal is to integrate structural information from numerous images into a single, high-accuracy three-dimensional reconstruction of the synaptic cytoskeleton. The immediate result of this collaboration will be an improved understanding of the spatial relationships between synaptic macromolecules. More importantly, the project will produce a set of computational tools that can be applied to stereo image data sets of various areas of the nervous system, from the macroscopic to the molecular level. Finally, our studies will advance the state-of-the-art of parallel c omputation and interactive reconstruction methods that can provide novel solutions to difficult problems of neuroscience visualization.
Showing the most recent 10 out of 384 publications
