Cluster-based nonparametric t-tests in Brainstorm

Hi I was reading from the brainstorm log page the following:

Some FieldTrip functions can be called from the pipeline editor in Brainstorm. This brings a lot of long-awaited tools for statistical analysis:

Permutation-based non-parametric t-tests
Cluster-based correction for multiple comparisons
This is done with direct calls to FieldTrip functions:
ft_timelockstatistics, ft_sourcestatistics, ft_freqstatistics
New displays for exploring significant clusters

I immediately updated the brainstorm on my pc to the latest version (Oct 9, 2015). However, in the pipeline editor I was not able to find any of the above tools… Any idea why? Your response will be highly appreciated.

Just an update. I did see the fieldtrip option for the two-sample t-test, but is it possible to do the nonparametric cluster test for the one-sample case?

Not yet, but it will be added at some point.

[QUOTE=Francois;9437]Not yet, but it will be added at some point.[/QUOTE]

For now, do you think it makes sense to create a source map file with zero amplitudes (suppose the null hypothesis is mean equal to zero) and use the paired test to achieve one-sample test?

I think it would be easier for you to add the missing processes, or help me do it.
I guess there are not many adaptations to do from the existing processes (process_ft_sourcestatistics.m, process_ft_timelockedstatistics.m, process_ft_timefreqstatistics.m).

If you are interested in learning how the programming of processes (=plugins) works, you can read the following page:
http://neuroimage.usc.edu/brainstorm/Tutorials/TutUserProcess

Alternatively, would you be able to tell me what needs to be changed in the FieldTrip calls in these existing processes?

Thanks,
Francois

Hi

Just checking in to see if the cluster-based one-sample test had been developed yet

Thanks

Nothing new on this front.
Do you have any implementation suggestions to share?

Unfortunately I do not. Will See what I can do. Thanks