Deep Sources Bug

Bugs
1

Hi everyone!
I just computed surface and deep structures sources for an EEG file, I followed the BS Tutorial for that purpose but, In last step, when everything is done, the display shows only sources for Right Hemisphere (and actually, the pathologic eeg activity comes from Left side in this recording)!

subcort
subcort906×567 93.5 KB

Steps I followed:

I'm using default anatomy
Followed every step in order to mesh surface and deep structures (amygdala and hipocampus),
I set the electrode positions
EEG recording is short, (a 8 seconds recording that I imported in database)
Computed noise covariance
Computed head model (OpenMeeg Mixed)
Computed Sources 2018 (Current Density Map)

I check again the anatomy mesh and everything is ok,

Help

Thanks in advance!

ps: I also had this issue with another EEG file, that time I juts switch the compute sources method to dSPM and it fixed it, why is that?

2

It looks like you didn't use the same constraints for the left and right hemisphere.

I'd recommend you use simpler source models instead: surface (as in the introduction tutorials) or volume (https://neuroimage.usc.edu/brainstorm/Tutorials/TutVolSource). These mixed head models are very complicated to deal with.

3

Thanks for the reply!

I actually juts followed the Tutorial steps, I made a screenshot, is everything ok in this mixed model?

deep
deep1089×429 62.5 KB

Thanks in advance!
by the way EEG file has only the 19 standard electrodes

ps: I tried with dSPM and Loreta and it works fine, problem is juts with Current Density Map option, but I'm not sure about using S-Loreta for Deep Structures since Loreta analysis was ment to be cortical, any suggestions?
:slight_smile:

4

is everything ok in this mixed model?

It looks OK, yes.

by the way EEG file has only the 19 standard electrodes

You can't expect much from minimum norm source modeling with 19 EEG electrodes. Nothing beyond a lobe-level precision can be expected. As I mentioned before, using this mixed head model is probably a lot of added complexity for no gain in precision.

I'm not sure about using sLORETA for Deep Structures since Loreta analysis was ment to be cortical, any suggestions?

@Sylvain @John_Mosher?

1 Like
5

I agree with François that 19 electrodes will yield very limited spatial resolution. As for deeper brain activity, our experience has been very positive with using the minimum-norm estimator of Brainstorm, followed by rectification of time sources and z-scoring with respect to a baseline reference.

6

Thank you!
I will try it!

7

Update:
I tried it again but this time with a 32 channel EEG file and I got the same issue: in cortical activations display , just sources for one hemisphere are shown, however, that was just when I computed head model with OpenMeeg, if I use Spheric (3-Shell) model, everything works fine (???)

Maybe is some head model specific issue..?

Thanks in adavance!

:thinking: