Abstract and subjects
MEG is a noninvasive measure of electrophysiological brain activity which provides excellent temporal and high spatial resolution. Because of its uniquely high temporal resolution relative to the more commonly used hemodynamic-based measures (fMRI, PET), the usefulness of MEG as a complementary neuroimaging method is becoming more widely recognized, particularly in the investigation of functional connectivity within and between large-scale brain networks. However, the available analysis methods for solving the inverse problem for MEG have yet to be compared and standardized. A comparison of analysis methods is further complicated by the fact that the different MEG systems have different data formats, noise cancellation methods, and sensor configurations. In order to facilitate this process, we established a website containing an extensive series of realistic simulated data for testing purposes (http://cobre.mrn.org/megsim/). In addition, we assert the usefulness of these datasets for training purposes, as they will provide an unambiguous answer to whether a trainee is correctly carrying out analyses. Here, we present a brief rationale and description of the testbed created, including cases emphasizing functional connectivity (e.g., oscillatory activity) and the default mode network (DMN). They are suitable for use with a wide assortment of analyses including equivalent current dipole (ECD), minimum norm, beamformers, independent component analysis (ICA), Granger causality/directed transfer function, and single-trial methods.