Computing Realistic Surrogate EEG for the Study of Functional Connectivity

Functional connectivity computed from electroencephalograms (EEG) can be used to better understand how the brain works. Unfortunately, estimating such connectivity is fraught with many pitfalls and can be confounded with artifacts due to volume conduction, common sources, reference scheme, etc. Devising a method to compute surrogate EEG that would be free of functional connectivity but that would reliably reproduce the effect of confounders such as volume conduction would be invaluable for statistical inference on functional connectivity. We developed such a method by simulating EEG from estimated sources and by reproducing the properties of local (but not long-range) functional connectivity in intracranial recordings. We present an example of how this approach can be used to improve the estimation of functional connectivity in EEG.