[FieldTrip] Extremely low power for the estimated sources

[Motaghi, P. (Pouria)]

Dear FieldTrippers,

I hope that you are well.

For my project, I am working on source reconstruction of EEG data. I have decided to proceed with minimum norm estimates. I do not have individual MRI scans so I am using a template head model. I have run into a problem and that is the output source powers are extremely low, meaning that the maximum has turned out to be around 1.75e-49. Everything (including data amplitudes, noise covariance, headmodel conductivities and leadfield magnitude) looks good until before calculating the inverse solution. I was wondering if this problem might have been caused by the rank deficiency problem since I have used a common average rereferencing in the preprocessing pipeline and I have also used ICA for removing eye and sometimes heart beat artifacts. Here is the code I used for solving the inverse solution:

        cfg                    = [];
        cfg.method             = 'mne';
        cfg.sourcemodel        = leadfield;
        cfg.headmodel          = vol;                                         %it is actually a field from the standard headmodel ‘standard_bem.mat’
        cfg.channel            = eeg_channels;
        cfg.mne.prewhiten      = 'yes';
        cfg.mne.lambda         = 3;
        cfg.mne.scalesourcecov = 'yes';

        source = ft_sourceanalysis(cfg, data_cond);

as you can see I have used prewhitening and lamda in the code but still when I run fprintf('Max abs source.avg.pow: %.6g\n', max(abs(source.avg.pow(:)))) I receive a value of around 1.75e-49 which I believe is too small. For this participant, I checked the covariance matrix’s rank and it was 56 out 60 (I have used 60 channels) which is aligned with the rejection of three ICA components and one for the common average reference. I was wondering if the problem arises from the rank deficiency. Do you think this might be the case? And if yes what should I do to correct for that? Do I have to run a separate prewhitening to compensate for the rank deficiency?

Thank you in advance for your response.

Kind regards/Met vriendelijke groeten,

Pouria Motaghi, M.D. | PhD Candidate | Behavioural Science Institute | Radboud University

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20260731/42ef309f/attachment.htm>