Hello,
I have some questions about the Melodic procedure via Temporal Concatenation and the subsequent dual regression method.
Our data is resting state data.
After running Melodic group analysis, there is a Subjects/Session mode graph underneath each component result. I read where this is described as “effect size”. Is this stating how strong each subject is contributing/driving the respective component? How is the effect size calculated at the individual subject level? What is the actual unit measure on the y-axis?
The GLM(OLS) on subject/session mode table reports betas, F-tests, and contrasts. Is the PE(1) the average effect size across subjects? What does the F test on the full model and contrasts indicate?
Regarding dual-regression, we had 36 components from our temporal concatenated ICA group analysis. Results from dual regression appear to provide a parameter-estimates file containing 36 components and a file of z-scores for each individual subject. However, I assumed that the z-stats file would also provide a z-score for each component, but it appears there are 72 subrick files instead of 36. I’m not sure why that is the case.
Your help would greatly be appreciated.
Thanks,
Michael
|