[FieldTrip] Baseline Correction using 'relchange' or 'db'

Stephen Whitmarsh stephen.whitmarsh at gmail.com
Wed Jul 29 09:26:34 CEST 2015


Dear Mick,

How did the designmatrix for ft_freqstatistics look? Maybe something went
wrong there.

Cheers,
Stephen

On 27 July 2015 at 18:24, <mick.lehmann at uzh.ch> wrote:

> Dear Fieldtrippers,
>
> I did the baseline correction using the following script:
>
>     TF_Remem_low = ft_freqbaseline(cfg, TF_Remem)
>     cfg.baseline     = [-1 -.1 ];
>     cfg.baselinetype = 'relchange'; HERE, I'VE TRIED 'DB' AND 'RELCHANGE',
> THE OUTCOME DIDN'T DIFFER
>     cfg.parameter    = 'powspctrm';
>
> This was done after ft_freqanalysis and before calculating the grand
> average.
>
> During the trial definition, I've already specified the baseline window
> and used 'demean':
>
> cfg.demean = 'yes';
> cfg.baselinewindow = [-1 -.1];
>
> When I made the figure for a contrast, it looked fine and exactly as I
> expected (see attachment). However, calculating the statistics using the
> cluster-based permutation test over a time window of 0-2000ms and between 5
> and 7 Hz (without averaging for time and frequency range), the comparison
> between the two conditions for every single time and frequency bin was
> highly significant (p< 0.001). This observation is not congruent with the
> figure, especially the fact that the t-values are almost equal across the 2
> seconds and across the freq range.
>
> Does anybody have an idea what might be the problem? Are the provided
> details sufficient for understanding the procedure?
>
> Here the script for the cluster-based permutation:
>
> cfg = [];
> cfg.channel          = 'all';
> cfg.latency          = [0 2];
> cfg.avgovertime      = 'no';
> cfg.frequency        = [5 7];
> cfg.avgoverfreq      = 'no';
> cfg.method           = 'montecarlo';
> cfg.statistic        = 'ft_statfun_depsamplesT';
> cfg.correctm         = 'cluster';
> cfg.clusteralpha     = 0.001;
> cfg.clusterstatistic = 'maxsum';
> cfg.minnbchan        = 2;
> cfg.tail             = 0;
> cfg.clustertail      = 0;
> cfg.alpha            = 0.001;
> cfg.numrandomization = 1000;
> cfg.neighbours       = neighbours;
>
>
> I'm deeply grateful for every comment.
> Best regards,
> Mick
>
> _______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20150729/3cfcc0a5/attachment-0002.html>


More information about the fieldtrip mailing list