[FieldTrip] Highlight significant values from ft_freqstatistics

Bruno Mansur brunommansur at gmail.com
Fri Jun 11 11:29:34 CEST 2021

Dear list members,

I've performed a cluster-based permutation test comparing time-frequency
data from two groups, each with 33 subj. For the perm. test I determined:

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.correcttail          = 'alpha';
cfg.alpha                 = 0.05;
cfg.numrandomization = 500;

So I guess, only *clusters with p<0.025 should be selected*.
Now I want to plot the results and highlight channels with p<0.025. For so,
I'm using *ft_topoplotTFR*:

neg_int        = zeros(numel(contrast.label),1);
neg_int(i1)    = all(neg(i2, m(k):m(k+1)), 2);
cfg.highlight  = 'on';
cfg.highlightchannel = find(*neg_int*);
cfg.layout     = 'acticap-64ch-standard2.mat';
ft_topoplotTFR(cfg, contrast);

According to '*neg_clusters_pvals*', the test has found *5 clusters* and *2*
of them are *significant* (*0.006* and *0.024*). In '*stat.prob*' and '
*stat.negclusterslabelmat*' I can see that the two significant clusters are
located between the *1st* and *5th* time-points (first clusters) and
between the *10th* and *11th* time-points. *However*, my *plot* shows one
big significant cluster from the *6th* to the *14th* time-points and one
small cluster at the *19th* time-point.

*Do you know why my plot differs so strongly from my data?* Am I
misunderstanding something?

I would really appreciate any help and I'd be happy to provide you further

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20210611/ac851853/attachment.htm>

More information about the fieldtrip mailing list