[FieldTrip] EEG source localization beamformer

[Schoffelen, J.M. (Jan Mathijs)]

Hi Sina,

There are 2 well-known issues related to beamforming EEG data that you need to look into, and tackle:

- depending on the type of volume conductor used, there might be numerical instabilities close to the edges/boundaries of the said volume conductor.
- beamformers have an inherent depth bias, which needs to be addressed in some kind of contrast/NAI etc. before any visualized result kan be interpreted in a meaningful way.

Please consult the literature (e.g. Britta’s paper:  https://www.sciencedirect.com/science/article/pii/S1053811921010612<https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.sciencedirect.com%2Fscience%2Farticle%2Fpii%2FS1053811921010612&data=05%7C02%7Cfieldtrip%40science.ru.nl%7C6537ec6e7e5a49d13e2108dcfca359b3%7C084578d9400d4a5aa7c7e76ca47af400%7C1%7C0%7C638663025711972955%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=nax3Rq7y0m8nKS3a5k8jSePtkIktT8bDnj8KZwkxYH8%3D&reserved=0>) and online resources to get on track.

Best wishes,
Jan-Mathijs


On 3 Nov 2024, at 21:35, Sina Makhdoomi Kaviri via fieldtrip <fieldtrip at science.ru.nl> wrote:

To whom it may concern,
I am working with a dataset related to reach and grasp tasks involving
six distinct motor tasks. I have divided the data based on event
markers and am using DICS Beamforming for source localization. Here is
an overview of the analysis steps I am following:

I am using a multitaper method for frequency analysis (mtmfft with
DPSS tapers) to compute the power and cross-spectral density (CSD) for
frequencies ranging from 0 to 100 Hz. Subsequently, I perform DICS
beamforming on the frequency-domain data to localize the sources in
the brain.

After obtaining the source estimates, I visualize the results using
orthogonal views. The problem I'm encountering is that, instead of
locating the main sources in the motor cortex as expected for the
reach and grasp tasks, the beamforming results show activation in the
parietal and occipital lobes. This is inconsistent with the expected
motor activity. Could you please help me identify any issues with my
approach or the code that might be causing this discrepancy?

cfg = [];
cfg.method = 'mtmfft';
cfg.taper = 'dpss';
cfg.output = 'powandcsd';
cfg.foi = 0:5:100; % or when I use 18
cfg.tapsmofrq = 2;
powcsd_all = ft_freqanalysis(cfg, EEG_avg);
cfg = [];
cfg.method = 'dics';
cfg.frequency = 18;
cfg.grid = leadfield;
cfg.headmodel = headmodel_eeg;
cfg.channel = matching_channels;
cfg.senstype = 'EEG';
cfg.dics.keepfilter = 'yes';
cfg.dics.lambda = '15%';
source = ft_sourceanalysis(cfg, powcsd_all);
%% Plot DICS Beamforming Results
cfg = [];
cfg.parameter = 'avg.pow';
sourceint = ft_sourceinterpolate(cfg, source, mri_resliced);
cfg = [];
cfg.method = 'ortho';
cfg.funparameter = 'pow';
cfg.funcolorlim = 'zeromax';
ft_sourceplot(cfg, sourceint);


Best regards,
Sina Makhdoomi Kaviri
PhD Candidate & Graduate Research Assistant
Sensorimotor Control Lab/Vinjamuri Lab
Department of Computer Science and Electrical Engineering
University of Maryland, Baltimore County
<task4_trial1_subjectG01_BF.png>_______________________________________________
fieldtrip mailing list
https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
https://doi.org/10.1371/journal.pcbi.1002202

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20241104/0356bd20/attachment.htm>