Hi, I'm trying to see how much I can get for EEG source reconstruction without individual MRI or digitization. So I'm just using a standard BEM (open MEEG BEM) and the channel locations seem to be correctly transformed. But for very clear N1 and P2 sound-onset responses, I get all sources around the frontal poles.
I've computed noise/data covariance for [-500,0] ms and [0,300] ms, respectively. I've tried covariance computing with time-locked trials and average ones, but both resulted in very similar localizations. Any suggestions/ideas, please?
Your EEG doesn't look like "a clear N1 and P2" to me, because you have high values during the first 50ms of recordings. Something went wrong either with the recording or the pre-processing of the recordings (check the filters, maybe try filtering less, or not filtering at all).
Other than that, the scalp topography you show match the sources you obtain, there is probably nothing wrong at that level. Note that for minimum norm imaging you don't need a data covariance.
Thanks a lot to take time to look at the figures!
Yes, I admit that it doesn't look too typical because it was filtered at [2,20] Hz. For [0.5, 30] Hz and at Cz, the response looks more typical:
Perhaps this is because we haven't corrected for the eye-related artifacts yet. Or perhaps the cap was slightly off than the standard positions?
Anyways, thanks for commenting on the agreement between topography and source locations! At least, I'm using Brainstorm right, only the data I'm putting into.
Then the minimum norm approach is not capable of distinguishing between a bilateral auditory response and a frontal activity... It is something common in EEG: the recordings as so smooth at the surface of the head that many configurations of brain activity can lead to exactly the same scalp potentials. Especially if you don't have many electrodes lower than the ear (and even if you do, they contain often too much muscle activity to be used in source modeling). Source analysis in EEG is complicated...
