Hi Francois,
For visual ERP exploration I have scaled dSPM results by using the Scale averaged dSPM-function. I assume that if I want to compare visually these wMNE Z-score maps to scaled dSPM maps, I just multiply these wMNE Z-score maps with the square root of averaged trials (as is done with dSPM). Is this right?
This second question might be more complicated. In epilepsy tutorial it was said that noise covariance matrix should be calculated from recordings that do not contain epileptic activity. I understand this, if we are studying the origin and propagation of the epileptic spikes.
I have 20 epilepsy patients in my study and they are all diagnosed with drug resistant epilepsy. In baseline files there was no seizures, but there might be some constant epileptiformic activity (for example inter-ictal spikes) and at least many of these subjects have abnormal EEG.
How should I select the data in baseline file and the data for noise covariance calculations in a way, that I can minimize the effect of different abnormal brain activity between the subjects in group analysis? For example, if I have found ERP and I want to find out where is it originating from and if I assume that this abnormal brain activity has no effect on this ERP, how should I proceed to normalize subject averages (dSPM or wMNE) before group analysis (noise covariance and baseline selection)?
Now, if my assumption is wrong and if these ERP sources would be influenced by individual epileptic activity, would it be so, that I could not see this effect on source level, if I have used baseline data which contains epileptic activity? On the other hand, if I find normal EEG for baseline and for noise covariance, would it be then possible to see the effect of epileptic activity to ERP sources?
Sorry, that this last topic was so messy. It is because I don't understand the effects of noise de-whitening and normalization to source maps well enough.
Thank you in advance!