[FieldTrip] source statistics

[Lilla.Magyari at mpi.nl]

Dear Seb,

I just would like to add to Jan-Mathijs's reply that
there is an example code for how to use ft_sourcestatistics here:

http://fieldtrip.fcdonders.nl/example/source_statistics?s[]=source&s[]=statistics

Best,
Lilla



> Dear Seb,
>
> A few comments to your question:
> -I don't really remember why 'pcc' as a method shouldn't work, but I
> believe that a straightforward explanation why it is disabled is that
> once-upon-a-time the functionality you are trying to apply was designed
> for 'dics' as method (and as such originates from the pre-pcc era. I
> wonder whether it will work for cfg.method = 'dics'.
> -Even if it would work, it is a very computationally heavy method: data
> are shuffled across the conditions at the channel level and spatial
> filters are computed (for both conditions) for each randomization. When we
> implemented this method, back in the old days, we quickly abandoned it
> again because of the discrepancy between the computation time and the
> computational resources.
> -The functionality still being present (I still assume that it kind of
> works) is the consequence of us being quite devoted to backward
> compatibility, although I don't want to bet my money on it to work (for
> dics).
> -Nowadays, I would advise to do the randomisation testing at the source
> level, i.e. computing a single set of spatial filters common to both
> conditions (the so-called common filter approach), and then use
> ft_sourcestatistics for the randomization testing. To get started, you
> could have a look at:
> http://fieldtrip.fcdonders.nl/example/common_filters_in_beamforming?s[]=common&s[]=filter,
> although I notice that the final step (calling ft_sourcestatistics is not
> explained there).
>
>
> Cheers,
>
> JM
>
> On Oct 27, 2011, at 7:35 PM, miellet at psy.gla.ac.uk wrote:
>
>> Hello,
>> I'm trying to use ft_sourceanalysis with pcc. I specify the volume
>> conductor model, the sensor information and a grid with pre-computed
>> leadfields.
>> I tried to compute the statistics over the source parameters between two
>> conditions as indicated in the documentation (ft_sourceanalysis(cfg,
>> freqA, freqB) with randomization).
>> I get an error linked to the condition:
>> if nargin>2 && strcmp(cfg.prewhiten,'no')
>> I couldn't find any information about this cfg.prewhiten parameter.
>> this part of the code if commented by
>> % HACK: requires some extra defaults
>> % HACK: use some experimental code
>> Does anyone know if this part is functional yet please?
>> Thanks,
>> Sebastien
>>
>> ----------------------------------------------------------------
>>  This message was sent using the Web mail system for
>>  The University of Glasgow School of Psychology
>> ------------------------------------------------------------------
>>
>>
>> _______________________________________________
>> 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
> J.Schoffelen at donders.ru.nl
> Telephone: +31-24-3614793
>
> _______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip