[FieldTrip] Questions on Permutation Tests on Results of Time-frequency Analysis by Frequency Band

[u3564255 at connect.hku.hk]

Dear FIELDTRIP LIST,

I am running cluster-based permutation tests on the results of time-frequency analysis. My goal is to compare two conditions in terms of delta, theta, alpha, beta, and gamma.

I applied morlet wavelets to decompose the data, and obtained power at frequencies from 1 Hz to 100 Hz. And then, I applied cluster-based permutation tests on the tf results, namely the power of all frequencies from 1 Hz to 100 Hz. As the neighbouring channel cluster file for the cap I use Quikcap 32 channel is not available online, I opted for the method "triangulation". (Could you please offer me a template file for Quikcap 32 channel if you happen to have one? Thank you!)

The permutation tests spitted out some negative and positive clusters reaching a significant level. I notice that one cluster includes basically all channels, except one. When I delve into stat.negclusterslabelmat (since it is a negative cluster), I realise that in fact within some frequency bands, only a subset of electrodes are involved in the significant difference between the two conditions. For example, in my data, only some central, frontal, and parietal electrodes are found to be significant in the delta band (1 - 2 Hz) in a particular time window.

Therefore, I would like to ask two questions:::

> First, could I say that the cluster for the delta band in my data is formed by that subset of central, frontal, and parietal electrodes instead of all electrodes except one?

> Second, should I run permutation tests for each frequency band (i.e., delta, theta, alpha, beta, and gamma) separately instead of running one permutation test on all frequencies from 1 Hz to 100 Hz? If I do so, will I run into the multiple comparison problem?

Thank you so much in advance!

Best regards,
Samuel

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20230528/4eb51250/attachment.htm>