Dear FSL Experts,
I had a confusion regarding input to dual regression regarding resting state data in a whole brain analysis.
Suppose we have 10 subjects and 10 controls, and we want to run dual regression on the file generated by MELODIC.
since melodic_IC being an input to dual regression it contains for eg. 42 components.
If we provide this melodic_IC, design.mat and design.con file to dual regression along with their the corresponding folder address for each subject and performing 5000 permutation using "randomize".
Provided design.mat and design.con file contains the information of grouping on the basis of subjects and control, the result (*corrp*) file shows the differences within the groups as per already defined contrasts for all 42 components.
As it is a two group contrast both ways for each RSN, If my *corrp* file still have some synchronous activity at p-values <0.05/42X2=0.0005 then the results survived are significant and can be reported .
Since I don't find any significant voxel at a pvalue <0.00005 I want to perform similar dual regression for 4 component so that I have to look for <0.05/4X2 =0.006
How to perform dual regression for 4 components or for eg. n components?
because I don't know where to do a change in the current procedure which I am following.
|