Hi all,
I think that perhaps it might be more natural to compute MNE solutions in the frequency domain using the noise cross-spectral density matrices rather than the time-domain noise covariance matrix.
The rest of the matrices used to compute the whitened MNE inverse operator would stay the same. Interestingly the operator will be frequency specific and complex! That might be a problem. For localizing frequency bands, averaging the noise cross-spectral density matrices across frequency bands might do the trick.
Rey.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20091202/f21959d6/attachment.html