[FieldTrip] Time-Frequency Plot Granger Causality

"Jörn M. Horschig" jm.horschig at donders.ru.nl
Thu Dec 12 09:39:50 CET 2013

Hi Lisandro,

actually, for me it works by just computing a time-frequency 
representation and shoving that into ft_connectivityanalysis. Is there 
an error if you try that? I literally do

     cfg = [];
     cfg.channelcmb = ...;
     cfg.method    = 'granger';
     granger    = ft_connectivityanalysis(cfg, freq);

With freq obtained by

       cfg           = [];
       cfg.method    = 'mtmconvol';
       cfg.output    = 'fourier';
       cfg.taper     = 'hanning';
       cfg.foi       = 1:32;
       cfg.t_ftimwin = 0.5.*ones(size(4./cfg.foi'), 1);
       cfg.toi       = -1:0.05:1;
       freq          = ft_freqanalysis(cfg, data);

Otherwise, solution 2 is something that definitely works and what I also 
did first, and it worked fine. There should be nothing wrong imho, as JM 
also pointed out.


Lisandro Kaunitz wrote:
> Hi,
> Following the tutorial on connectivity analysis I am trying to 
> implement Granger Causality on simulated signals. I can reproduce all 
> the results that appear in the tutorial in my dataset but I cannot 
> obtain a graph of Wavelet-based granger causality with the format:
> -x axis: time
> -y axis: frequency
> This  graph should allow me to compare in a single figure how channel 
> 1 granger causes channel 2 at a range of frequencies as a function of 
> time (like in Fig 3 of Dhamala et al, 2008).
> My problem is that I cannot figure out how to produce this graph with 
> the tutorial functions as they seem to automatically compute Granger 
> Spectra for the whole epoch. In the tutorial demo the structure 'freq' 
> has dimord: 'rpttap_chan_freq', and the structure 'granger' has 
> dimord: 'chan_chan_freq'. What I am looking for is a function that 
> returns a structure with the granger spectra computed with dimord of 
> 'chan_chan_freq_time'
> 1- I've tried running 'freq = ft_freqanalysis(cfg, data);' and then 
> '[output] = ft_connectivity_csd2transfer(freq);' with all types of 
> parameters but I could not reach to a solution. Is there a way of 
> building a time-frequency granger spectra within fieldtrip?
> 2- Alternatively to 1, I thought that I could compute non-parametric 
> Granger Spectra for different time windows inside each epoch and then 
> concatenate the results. Before doing this I would like to be sure 
> that I am not missing a way of doing it that it's already implemented 
> in fieldtrip (I have just started using Fieldtrip a short time ago)
> Thanks a lot
> Cheers
> Lisandro Kaunitz
> _______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip

Jörn M. Horschig
PhD Student
Donders Institute for Brain, Cognition and Behaviour
Centre for Cognitive Neuroimaging
Radboud University Nijmegen
Neuronal Oscillations Group
FieldTrip Development Team

P.O. Box 9101
NL-6500 HB Nijmegen
The Netherlands

E-Mail: jm.horschig at donders.ru.nl
Tel:    +31-(0)24-36-68493
Web: http://www.ru.nl/donders

Visiting address:
Trigon, room 2.30
Kapittelweg 29
NL-6525 EN Nijmegen
The Netherlands

More information about the fieldtrip mailing list