Only about a decade or so ago, the cost of terabyte level data storage was in the million dollar range;currently, it has fallen dramatically to a few hundred dollars. This has opened up unprecedented opportunities to make long term, continuous recordings of neural activity using microelectrode arrays, EEG, MEG and other methods. This will permit quantitative study of brain activity over an enormous range of time scales ranging from milliseconds to years. However, we are lacking adequate software tools to effectively manage, visualize, browse and analyze such large data sets of neural signals. In this grant proposal we intend to carry out the research needed to directly address this shortcoming through enhancements to the Chronux platform for neural signal processing. The Chronux platform currently exists as a MATLAB toolbox, implementing sophisticated signal processing algorithms for both continuous (LFP, EEG) and point (spike times) process data in single or multiple channels. The toolbox has gained widespread acceptance in the neuroscience community, with currently over 3700 downloads and 374 voluntary registrations from users, citations in numerous scientific papers, and pedagogical usage in courses. Here we propose to enhance Chronux functionality, allowing the handling of multi-terabyte data sets, and increasing usability by basic and clinical researchers.
Specific aims i nclude enhancements to core routines to permit efficient data IO and computations, development of a client-server architecture and web interface, and a data management system for large neural recording sets. The innovations required by this research proposal lie at the interface between signal processing, computer science and quantitative neuroscience. If successful, the results will significantly enhance the ability of basic and clinical researchers to perform sophisticated, multi-scale analysis of large neural recordings, and permit the analysis of abnormal patterns of brain activity for development of clinical indicators or evaluation of therapeutic interventions.
Exponential decline in storage prices have made it possible to acquire, at reasonable costs, very large volumes of recordings (terabytes to petabytes) of neural activity using microelectrodes, EEG, MEG and fMRI. Currently, we lack powerful tools to visualize and analyze datasets of such large magnitude. Neurological and neuropsychiatric disorders are associated with abnormal patterns of brain activity, and having effective tools to visualize, quantify and analyze such patterns within very large datasets are important for both basic and clinical research. This grant proposal is to enhance the Chronux software platform for neural signal processing, already established in the neuroscience research community, to be able to handle such large datasets in a user friendly way, so that basic and clinical researchers can more effectively study normal and pathological brain dynamics, and develop clinical indicators for diagnosis and therapeutic development.
