[Eeglablist] Not timefrequency data on fieldtrip: cluster-based statistics

[Ramtin Mehraram (Student)]

Hi all!

This question relates to the Fieldtrip toolbox, but I hope anyone can still help me.

I need to run a statistical analysis across four independent groups for some network measure I obtained with the brain connectivity toolbox (degree, clustering coefficient, etc.) from EEG data. I am dealing with the multiple comparison issue. Since I am using a high density cap, nothing survives the Bonferroni correction (and variants), nor the FDR.

I am trying to perform a cluster-based (montecarlo) analysis as implemented in Fieldtrip, but the toolbox needs the input to be organized as a time-frequency structure. I tried to rebuild a similar structure (degree values instead of power spectrum values), but it seems that I am being unsuccessful.

Indeed: if I run a "cfg.statistic = indepsamplesF" across the groups I get NaNs in the probability output matrix. If I just perform a "cfg.statistic = indepsamplesT" between two random groups, I get a couple of channels or so with p<0.05.

In all the cases, any form of fieldtrip plotting (such as ft_clusterplot) shows as output the message "Argument to mbscalar must be scalar".

Any idea? Is it actually sensible to proceed in this way or is there any smarter way to work with not time-frequency data on fieldtrip?

Many thanks in advance for your help.

Ramtin Mehraram
PhD Student @ramtinTVT
Biomedical Research Building 3rd floor
Institute of Neuroscience
Newcastle University
NE4 5PL, United Kingdom
www.lewybodylab.org
https://www.newcastlebrc.nihr.ac.uk/research-themes/dementia/