[FieldTrip] Source reconstruction from EEG 64 electrodes

[Emilie Caspar]

Dear community, 

I have several questions regarding source localization. Basically I have never done source localization with EEG so the questions may be a bit broad. 

I have conducted two similar experiments, one in MRI and one in EEG (with Biosemi, different participants). Basically we observe differences in two experimental conditions in EEG but not between the two same conditions in MRI. We thus would like to localize the source of the differences between the ERPs in EEG, and then to apply this to MRI data. The EEG has 64 electrodes (classic headcap) and we do not have MRI anatomical scan for the EEG participants. 

Questions:
(1) I don’t have anatomical scans for EEG participants, I was wondering what was the best option to select a standard head model file. (I have found this standard_mi.mat file here: https://github.com/fieldtrip/fieldtrip/blob/master/template/headmodel/standard_mri.mat <https://github.com/fieldtrip/fieldtrip/blob/master/template/headmodel/standard_mri.mat>) but I don’t know if it’s the best head model for my set up or not. Would you have some recommendations?

(2)  I have also tried to start a script for source localization. I first timelock the data as follow, after preprocessing, which works well.
        cfg = [];
        cfg.trials = 'all';
        cfg.covariance = 'yes';
        cfg.covariancewindow = [-inf 0]; 
        AVG_ERP_Shock_Source = ft_timelockanalysis(cfg, cleandata);
        
And then I calculate the inverse solution:

vol = ft_read_headmodel('/Volumes/Commander_Empathy_EEG/standard_mri.mat');
 
cfg               = [];
cfg.method        = 'mne';
cfg.grid          = 'biosemi64.lay';
cfg.headmodel     = vol;
cfg.mne.prewhiten = 'yes';
cfg.mne.lambda    = 3;
cfg.mne.scalesourcecov = 'yes';
sourceSHOCK          = ft_sourceanalysis(cfg, AVG_ERP_Shock_Source);

But here ft_read_headmodel is not recognized as a function (Undefined function or variable ‘ft_read_headmodel’).
Would you know what is the issue?

(3) For cfg.grid, I used the classical electrode headcap from biosemi so I though that using the biosemi64.lay would be good since it contains the position of each electrode, but I am not sure at all that this is the case. Would someone know more about this?

(4) And then a final question, is is better to compute a source localization for each participant individually and then to average them, or to perform the source localization on the grand average? 

Many thanks for your help!!

Emilie



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20191129/2d935af2/attachment.html>