[FieldTrip] Beamforming on resting state data

Tobias Navarro Schroeder tobiasnavarroschroeder at googlemail.com
Tue May 31 14:12:48 CEST 2011


Hi Michael,

thanks a lot for that suggestion.
The gamma activity at the sensor level might indeed be of muscular origin
and this would explain why it doesn't show up inside the head.
Even after automatic & manual artifact rejection the patients were left with
more gamma activity. To my 'untrained eye' the sensor topographies suggested
that the activity would come from the brain. It didn't extend to the
boundaries of the Topoplot, at least in the 60-90 Hz range. However, this
was the case for other parts of the gamma-band.

Greetings, Tobias


On Tue, May 31, 2011 at 12:44 PM, Michael Wibral <michael.wibral at web.de>wrote:

> Hi Tobias,
>
> it could be that the temporal activity at the sensor level is of muscular
> origin. Given all runs well in beamforming you shouldn't see too much of
> that insode the head - have you considered this possibility?
>
> Michael
>
>
>
>  ------------------------------
> *Von:* "Tobias Navarro Schroeder" <tobiasnavarroschroeder at googlemail.com>
> *Gesendet:* May 30, 2011 5:58:32 PM
> *An:* fieldtrip at donders.ru.nl
> *Betreff:* [FieldTrip] Beamforming on resting state data
>
>
>  Hi Fieldtrippers,
>
> I performed DICS beamforming on resting-state data (i.e. no task, no
> baseline, eyes closed) of a clinical population and controls.
> According to the sensor data the patients have more gamma-band (60-90
> Hz) activity over temporal areas (somewhat spread-out).
> Curiously, after beamforming and group comparisons with 'indepsamplesT'
> permutation statistics the difference map shows more gamma-band (60-90
> Hz) activity for controls in the left, middle-temporal gyrus.
> Hence, the source-level results are opposite to the sensor-level results
> (although more spatially confined). This is *not* a problem of the design
> matrix, or confusing the groups.
>
> Question 1: Does anybody have experience with opposite results on source
> and sensor level?
>
> The individual raw beamforming results (interpolated on the subjects' MRIs)
> look suspicios as well. Due to the 'unit gain constraint' there should be an
> overestimation of power in the center of the head, but only a minority of
> subjects show that. Often I see temporal or parietal activity.
>
> Question 2:
>
> Would a lack of a blob in the center of the head in individual,
> raw, beamforming data indicate an erroneous source analysis?
>
> Thanks a lot for help and suggestions.
>
> Best, Tobias Navarro Schröder
>
> ps: The configurations for calculating the covariance and the beamforming
> is added below:
>
>
>
> 1) % Compute the covariance matrices
>
>
>
> cfg.output =
> 'powandcsd';
>
> cfg.foi = 75;
>
> cfg.foilim = [75 75];
>
> cfg.tapsmofrq = 15; % 75 +- 15 = 60 - 90 Hz
>
> cfg.pad =
> 'maxperlen'; % default
>
> cfg.method =
> 'wltconvol'; % wavelet analysis
>
> cfg.width = 5;
> % Width of the wavelet. Determines the temporal and spectral resolution of
> the analysis constant.
>
>
>
> 2) % Perform the actual beamforming
>
> % load grids of MRI volume
>
> fullnameGrid = [ CurSubject.mripath CurSubject.MRIModel];
>
> cfg = load(fullnameGrid);
>
> cfg.vol = cfg.hdm;
>
>
>
>
>
> cfg.frequency = 75;
>
> cfg.method =
> 'dics';
>
> cfg.keepfilter =
> 'yes'; % keep the beamformer filters (virtual electrodes) for later use
>
> cfg.keeptrial =
> 'no';
>
> cfg.keepmom =
> 'yes'; % useful for connectivity calculations. Dipole moments..
>
> cfg.projectmom =
> 'yes';
>
> cfg.projectnoise =
> 'yes';
>
> cfg.latency = 1;
>
>
> _______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20110531/746845e2/attachment.html>


More information about the fieldtrip mailing list