[FieldTrip] covariance computation from ft_timelockanalysis

[Luke Bloy]

Dear Jan-Mathijs

Thank you for the quick response. Computing the covariance seems to be
working fine now.

I'm curious what you would suggest for computing the differential power?

at the moment I'm not keeping any trial measures (cfg.keeptrials='no') in
either ft_timelockanalysis or ft_sourceanalysis. I'm then computing a
psuedoT as

act_sourceAnalysis.avg.pseudoT = (act_sourceAnalysis.avg.pow -
cnt_sourceAnalysis.avg.pow)./(act_sourceAnalysis.avg.noise +
cnt_sourceAnalysis.avg.noise);

I wanted to compare this pseudoT to a standard T score.

I can get trial by trial power estimates by using cfg.keeptrials='yes'
in ft_timelockanalysis and  cfg.keeptrials='yes' and ctf.rawtrial='yes' in
ft_sourceanalysis. but ft_sourcestatistics doesn't seem to let me do a
paired T the way I would hope. Is there an alternative route that you might
take or is there another method that could be useful?

Thanks,
Luke







On Thu, Apr 17, 2014 at 4:27 AM, jan-mathijs schoffelen <
jan.schoffelen at donders.ru.nl> wrote:

> Hi Luke,
>
> On Apr 16, 2014, at 9:29 PM, Luke Bloy wrote:
>
> Hi All,
>
> I'm interested in using lcmv to do a common mode differential beamformer.
> My understanding, please correct me if I'm incorrect, is that the best way
> forward would be to
> 1) compute a covariance matrix from a time window combining both active
> and control windows.
> 2) use this combined covariance to make a common mode lcmv beamformer
> 3) apply this to both active and control windows separately
> 4) combine into some sort of psuedo T or NAI.
>
> does this sound correct?
>
>
> Yes it does.
>
> If so the main thing I'm unclear on is how to compute the covariance for 2
> distinct time intervals? is this possible?
>
>
> What you could do is call ft_timelockanalysis twice: each time with a
> different cfg.covariancewindow. Assuming you have an equal number of trials
> for both time windows, averaging the two covariance matrices will give you
> the combined covariance matrix.
>
> Best,
> Jan-Mathijs
>
>
>
> Thanks,
> Luke
> _______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
>
>
>     Jan-Mathijs Schoffelen, MD PhD
>
> Donders Institute for Brain, Cognition and Behaviour,
> Centre for Cognitive Neuroimaging,
> Radboud University Nijmegen, The Netherlands
>
> Max Planck Institute for Psycholinguistics,
> Nijmegen, The Netherlands
>
> J.Schoffelen at donders.ru.nl
> Telephone: +31-24-3614793
>
> http://www.hettaligebrein.nl
>
>
> _______________________________________________
> 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/20140417/66fbeacf/attachment.html>