[FieldTrip] Subject level non parametric statistics for coherence with external stimulus

[Davide Tabarelli]

Dear Fieldtrip users,

I’m trying to calculate a statistical map for coherence differences between two conditions at a source level for a single subject, but I have some problems with channel combinations.

I have calculated common LCMV filters for my subject and computed source time series, that I have stored in a ft_timelock structure.
I have also successfully calculated coherence between the stimulus function and all dipoles for both conditions A and B using ft_freqanalysis and ft_connectivityanalysis.
Now I would like to compute a non parametric statistical map for the coherence difference between A and B using the approach of Maris & Schoffelen & Fries 2007.

I’m trying to use the “ft_statfun_indepsamplesZcoh” statistics as follow:

cfg=[]; 
cfg.parameter = 'fourierspctrm’; 
cfg.frequency = 5.5;
cfg.statistic = 'ft_statfun_indepsamplesZcoh’; 
cfg.method = 'montecarlo’; 
cfg.numrandomization = 1000;
cfg.design = design;
stat = ft_freqstatistics(icfg, fourier_conditionA, fourier_conditionA);

I realized this will compute the statistics for all the possible combination of channels, thus between stimulus and sources and between all pairs of sources … that is computationally non effordable (at least for me).

There is a way to tell ft_freqstatistics to use only some combination when computing the significance of coherence differences?

Or am I doing something wrong?

Thank you all !

D.

—
Davide Tabarelli, Ph.D.
Center for Mind Brain Sciences (CIMeC)
University of Trento,
Via delle Regole, 101
38123 Mattarello (TN)
Tel: +39 (0)461 283644
Italy