Comprehensive descriptions of pathology afforded by neuroimaging techniques require accurate reconstructions of both time and space. While hemodynamic techniques (fMRI and PET) have made significant contributions to health-related research by revealing the presence or absence of pathology in specific brain locations, pathology often exhibits itself first as a timing problem within and between brain regions (e.g., slowing or dysfunctional communication between brain areas). Magnetoencephalographic (MEG) methods can provide precise temporal information but the physics of this technique is complex and progress in this area has been slow. There are three primary manufacturers of the ~25 MEG systems in the United States, each of which has different hardware (pickup coils, sensor arrays, noise cancellation methods) and software, making it difficult for MEG researchers to: 1) directly compare simulation results testing their different analysis algorithms and 2) pool data across sites in order to acquire larger patient numbers. The goals of this project are as follows.
Specific Aim 1 requires the generation of a large set of simulated visual, somatosensory and auditory MEG data using realistic source configurations garnered from our basic/clinical studies, embedded within real background activity and formatted for the 3 different commercial MEG systems. Four general types of analysis techniques (multidipole, spatio-temporal modeling, L2 minimum- norm, L1 minimum-norm, and beamformer methods) will be applied to the simulated data.
Specific Aim 2 will test the same set of algorithms on empirical data (basic visual, auditory and somatosensory protocols) already acquired from three different MEG systems and from 5 subjects. The background MEG activity from the 5 subjects will be used in the simulations noted above.
Specific Aim 3 proposes to finalize modifications to free software that can read, display and analyze data from the three MEG systems and to write the simulated and empirical data into Curry (a commercial package) and netCDF formats so other investigator across the country and world may access these data and tools for direct comparison with their results. A secure collaborative portal will be set up and maintained at The MIND Institute for this purpose. MEG methods for examining brain function provide unparalleled information concerning the timing of activity within and across brain regions. Many brain pathologies disturb this timing, so that activity is either delayed or brain regions do not work well together in a concerted fashion. This project provides a comprehensive evaluation of varied MEG tools that capitalize on this strength. ? ? ?
| Aine, C J; Sanfratello, L; Ranken, D et al. (2012) MEG-SIM: a web portal for testing MEG analysis methods using realistic simulated and empirical data. Neuroinformatics 10:141-58 |
| Shoemaker, J M; Holdsworth, M T; Aine, C et al. (2011) A practical approach to incidental findings in neuroimaging research. Neurology 77:2123-7 |
