help > Regressing behavioural scores with ICA components
Showing 1-1 of 1 posts
Display:
Results per page:
Jul 5, 2019  03:07 PM | fmri questions
Regressing behavioural scores with ICA components
Dear All,

I want to perform a voxel level regression analysis for the DMN with some behavioural scores. I have done ICA analysis and identified the component number which mostly corresponds to the DMN. Now, I need to get subject level back projections corresponding to the component number and then do a voxel level regression within this DMN mask.

My questions are:

1) when I look at first level output, I have the files ICA_Measure001_*; I assume this is the group-ICA component 3; how do I threshold this image to get my DMN mask? Is it that everything above a value of 0 is component 3?

2) what's the difference between the ICA_Maps, ICA_ROIs, and the ICA_Measure* files?

3) Is it correct to take each subjects' specific component file as my y variable and the score as x variable and do a simple regression analysis in MATLAB or are there some other aspects that need to be accounted for? Can this analysis be done via GUI? I am interested in seeing how much of the component's beta value per voxel can my behavioral score explain.


Thanks


Best Regards
fmriquestions