Optimizing a Pipelined HPCC Processing Environment for Computational Neuroscience

University Of California Los Angeles, Los Angeles CA

Investigators

Abstract

DESCRIPTION (provided by applicant): Innovations in imaging technologies and in the field of computational neuroscience have resulted in the generation of massive amounts of image and statistical data concerning the human brain. These advances have resulted in the need for high performance computing and novel investigative paradigms for the meaningful analysis of the available data. The Laboratory of Neuro Imaging has been a frontrunner in the adoption of cutting-edge technology to understand dynamic changes such as the development and degeneration of the human brain in health and disease. Our group has gained worldwide recognition for the development of innovative research methodologies, computational algorithms and high-order mathematical approaches to the investigation of brain registration, the analysis of variance between and within populations, and the visualization of these data. These techniques, however, now must not only accommodate four dimensions, as ongoing projects at LONI and elsewhere generate time-varying, multidimensional statistical fields but also be able to reasonably process data that are increasingly more complex. We now routinely apply these computationally demanding methods to population studies in order to achieve sensitivity to potentially subtle differences. In response to these computational challenges, a group of neuro-, biomedical and computer scientists with common interests and computational needs have come together to seek funding to modernize the network infrastructure of a shared, dedicated high performance computing cluster. The increasing computational demands placed on this system by four-dimensional volume computation, intricate surface extraction, nonlinear warping and the multi-modal integration of complex isosurfaces with data from disparate sources have clearly identified key network bottlenecks that significantly degrade the processing performance of complex analyses. The requested network upgrade would eliminate congestion and bandwidth contention and allow for the optimal utilization of the computational resource by LONI investigators and collaborators. An administrative plan is already in place by which the equipment can be managed equitably. Technical and management personnel also are part of the funded group of participants. Ongoing collaborations and the common programmatical requirements will enable sharing of computer code, analytic procedures, computational strategies and infrastructural capabilities. The provision of the requested instrument will enhance the productivity of ongoing computational neuroscience research at LONI and collaborating sites in schizophrenia, HIV/AIDS and Alzheimer's disease, among others, and foster the development of leading edge technology and applications for all participants. PUBLIC HEALTH RELEVANCE: The Laboratory of Neuro Imaging (LONI) at the University of California, Los Angeles (UCLA) seeks funding to augment the network infrastructure of a high performance computational cluster utilized by an extensive and distinguished group of local, national and international neuroscientists with a common dedication to the study of the complex, dynamic brain in health and disease. The requested instrumentation package will alleviate significant network congestion and bandwidth contention detrimental to current and future investigations requiring the utilization of this shared computational resource. If funded, ongoing neuroscience research at LONI and a compelling number of collaborative efforts involving outside institutions stand to directly benefit.