Computing source models by epoch or on the average?

Discussions
1

Hi Francois,

Is there a benefit (in terms of SNR) to calculating the sources by epoch vs. average first and then sources, if I’m only going to do group stats with them afterwards? (I ask because of course the former takes much longer!)

Emily

2

Hi Emily,

In one single acquisition run: MinNorm(Average(trials)) = Average(MinNorm(trials)).
This is valid is you select the option “Minimum norm (wMNE)” in the old interface, or “Minimum norm imaging / current density map” in the new one.

If you have multiple runs, you may have issues with the position of the head.
You can do the following for each run: compute the average in sensor space (average of the recordings), and use the min norm maps for this average.
Then average the sources for the average across runs.
And finally normalize (Z-score)

Francois

3

That’s good news, thanks François.