[FieldTrip] Common beamforming filters for different groups?

[Rojas, Don]

Ulrich,

It does not make sense to me to have a common filter per pair of subjects, only between conditions within a subject (which you say you are currently doing). Since, as Johanna suggested, you are using a standard brain and standard electrode locations, rather than individual ones, you can have a common leadfield matrix for all subjects. However, I don't think you can have a common beamformer filter, since that is computed for each subject based on the covariance or cross spectral density matrix, which would be unique to each subject. Therefore, the beamformer filter, which uses both the covariance and the leadfield, would be unique for each individual. You can save time simply by having a common leadfield matrix.

Best,

Don

-----------------------
Don Rojas, Ph.D.
Associate Professor of Psychiatry
U. of Colorado Denver Anschutz Medical Campus
Director, UCD Magnetoencephalography Lab
13001 E. 17th Pl F546
Aurora, CO 80045
303-724-4994

On Jul 6, 2012, at 7:21 AM, Pomper, Ulrich wrote:

> Dear Johanna,
> thanks for your quick reply!
> 
> To clarify things: This is an EEG study, using a standard MNI-based headmodel for all subjects.
> 
> I have a between subjects factor group (patients, controls) and a within subjects factor condition (A, B). Also, I can use a within subject contrast (active versus baseline) for the source analysis.
> 
> Currently, I use one common filter for baseline and active data as well as both conditions for each subject. So I end up using one common filter per subject.
> 
> My question I whether it would be more correct to additionally pool the data from matched subjects into one single filter. Thus resulting in one common filter per pair.
> 
> 
> Cheers, Ulrich
> 
> 
> 
> 
> 
> 
> 
> 
> 
> ________________________________________
> From: fieldtrip-bounces at science.ru.nl [fieldtrip-bounces at science.ru.nl] On Behalf Of Johanna Zumer [johanna.zumer at donders.ru.nl]
> Sent: Friday, July 06, 2012 2:36 PM
> To: FieldTrip discussion list
> Subject: Re: [FieldTrip] Common beamforming filters for different groups?
> 
> Dear Ulrich,
> 
> Only if the leadfields are the same, can you use the common filter.
> 
> If you have EEG data with standard electrode positions based on the cap layout (rather than measured electrode positions) and using the standard MNI MRI rather than subject specific MRI, then your leadfield will be the same for all subjects.  Then for the covariance computation you might need to consider some normalization (z-transform) of the data (using some control window) so that differences in sensor values due to e.g. sensitivity of electrodes across subjects, won't bias the result.
> 
> If you have MEG data, then the different head positions in the helmet will mean different lead fields.  The only ways around that might be: if you do something similar to correcting for head movement within a single session (something like ft_headmovement but as far as I know is not implemented for this multi-subject /session case) or if you ensured similar head position between subject-pairs using ft_realtime_headlocalizer at the time of acquisition; and in combination with using the MNI MRI rather than individual subject MRI.
> 
> Perhaps someone else on the list can comment as to which is better, if the common filter is a possibility? A common filter using sub-optimal reconstruction for each subject if based on non-subject-specific MRI etc, versus optimal reconstruction per subject but then issue of not a common filter?
> 
> Another thing to do is to use the subject-specific inverse filter and do a contrast within that subject (active versus baseline) and take this contrast to the comparison at the paired-subjects group level. However, not all experiments have the possibility of within-subject active-versus-baseline.
> 
> Cheers,
> Johanna
> 
> 
> 2012/7/6 Pomper, Ulrich <Ulrich.Pomper at charite.de<mailto:Ulrich.Pomper at charite.de>>
> Dear list members,
> I want to compare source data between two different groups (patients and controls). Each patient has an age and gender-matched control.
> Is it correct to compute a common filter for each pair of subjects? The TF-topographies do look different within each pair, so I would assume that the sources are somewhat different. However, as I'm going to statistically compare the two groups it would be nice to have common filters, as otherwise any finding could be due to different filters as opposed to different sources.
> 
> Any help with this issue is greatly aprreciated!
> Cheers, Ulrich
> _______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl<mailto:fieldtrip at donders.ru.nl>
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> 
> 
> --
> Pflichtangaben gemäß Gesetz über elektronische Handelsregister und Genossenschaftsregister sowie das Unternehmensregister (EHUG):
> 
> Universitätsklinikum Hamburg-Eppendorf; Körperschaft des öffentlichen Rechts; Gerichtsstand: Hamburg
> 
> Vorstandsmitglieder: Prof. Dr. Guido Sauter (Vertreter des Vorsitzenden), Dr. Alexander Kirstein, Joachim Prölß, Prof. Dr. Dr. Uwe Koch-Gromus
> 
> _______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip