help > ICA network (intermediate steps and results files)
Dec 6, 2018  02:12 PM | Che Liu - University of Miami Miller School of Medicine
ICA network (intermediate steps and results files)
Dear CONN community and Alfonso,

Excellent work and very friendly interface. Saved a lot timefor fMRI beginners like me.
I have a few questions regarding ICA analysis in CONN. I am trying to understand intermediate steps.

First, is threshold based on the z scores only for visualization of spatial map of group-level compoents? Does this threshold affect the timeseries for each component? I guess time series corresponds to group-ICA map before thresholding (i.e. threshold=0). And how do you deal with negative z-score? I was trying to identify DMN, however, even after thresholding, there are two clusters with negative z-score, which does not belong to DMN (number of IC=20 or 25). Any suggestion?

Second, does thresholding group-ICA map affect the subject-level ICA results? I guess the subject-level ICA results will be always the same for a certain group-ICA. Also you mentioned subject-ICA map (after backprojection) is saved in 'firstlevel V2V BETA_Subject*_Condition*_Measure*_Component*.nii', however, what is the value for each voxel in this volume? z-score (normalized factor score like in group-ICA) or connectivity? Is this *nii same as 'Results explorer plot subjects'