[FieldTrip] Cluster based permutation test

[Patrick Wiegel]

Dear Fieldtrip Community,

I would like to compare time-frequency data between 2 conditions: Correct and Incorrect movements. For this purpose, I performed time-frequency analyses (Morlet wavelet) and cluster based permutation tests. The scripts run fine but I do have two more general questions regarding the analyses.

1) For each subject, the 2 conditions have a different amount of trials for the time-frequency analyses. In some cases, the differences are relatively great (e.g. 30 trials vs. 140 trials). At the moment, I am taking the average from these two conditions (time-frequency data) for each subject and using these for the group level cluster based permutation tests. Based on the great differences in the number of trials, I am not sure if this is a valid approach. I considered doing bootstrapping of the trials to have the same number of trials for the time-frequency analyses (I did the same for the behavioral data). However, I did not find the option in fiedltrip and I would have do it manually in a loop. Does anyone have experiences with dealing with such an issue?

2) Before the group level cluster based permutation test, I am a-priori defining a time and frequency window (theta, alpha, beta, gamma) of interest using ft_select data. The cluster based permutation test reveals differences between the 2 conditions for some frequency bands (I attached the code and the cluster plot for the gamma band which seems to comprise frontal and parietal regions).

As documented in the tutorials, it is difficult to make inferences about the spatial locality of this effect with this test (https://www.fieldtriptoolbox.org/faq/how_not_to_interpret_results_from_a_cluster-based_permutation_test/). I was wondering on how to interpret and present the results with regard to the spatial domain. Does anyone recommend an additional test to strengthen the analyses in the spatial domain (e.g. source localization)? Is there a common/good approach following the cluster based permutation test?

Any advice would be appreciated.

Best, Patrick



%%
cfg = [];
% cfg.channel          = {'Fz','F1','F2','F3','F4','F5','F6','F7','F8','Fc1','Fc2','Fc3','Fc4','Fc5','Fc6','Fc7','Fc8','Af3','Afz','Af4','Af7','Af8'};
cfg.parameter        = 'powspctrm';
cfg.method           = 'montecarlo';
cfg.statistic        = 'ft_statfun_depsamplesT';
cfg.correctm         = 'cluster';
cfg.clusteralpha     = 0.05;
cfg.clusterstatistic = 'maxsum';
cfg.minnbchan        = 2;
cfg.tail             = 0;
cfg.clustertail      = 0;
cfg.alpha            = 0.05;
cfg.numrandomization = 1000;
cfg.neighbours       = neighbours;


subj = 25;
design = zeros(2,2*subj);
for i = 1:subj
  design(1,i) = i;
end
for i = 1:subj
  design(1,subj+i) = i;
end
design(2,1:subj)        = 1;
design(2,subj+1:2*subj) = 2;

cfg.design   = design;
cfg.uvar     = 1;
cfg.ivar     = 2;

[stat_beta] = ft_freqstatistics(cfg, grandavg_single_1_beta, grandavg_single_0_beta);
[stat_theta] = ft_freqstatistics(cfg, grandavg_single_1_theta, grandavg_single_0_theta);
[stat_alpha] = ft_freqstatistics(cfg, grandavg_single_1_alpha, grandavg_single_0_alpha);
[stat_gamma] = ft_freqstatistics(cfg, grandavg_single_1_gamma, grandavg_single_0_gamma);


[cid:D00B1A08-AD26-4508-B5A5-5ADDB39160E8 at w2k.sport.uni-freiburg.de]
Patrick Wiegel
Department of Sport and Sport Science

University of Freiburg
Sandfangweg 4
79117 Freiburg i. Br .

phone: +49 (0)761/ 203-4550
email: patrick.wiegel at sport.uni-freiburg.de<mailto:patrick.wiegel at sport.uni-freiburg.de>
web: www.sport.uni-freiburg.de

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20201022/e18a41c8/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: gamma_cluster.jpg
Type: image/jpeg
Size: 7201 bytes
Desc: gamma_cluster.jpg
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20201022/e18a41c8/attachment.jpg>