Hello,

I am looking into the possibility of using ICA AROMA in addition to regressing white matter and CSF signals and temporal filtering with CONN. I am planning on conducting a coactivation pattern analysis on resting state data, where scrubbing could distort the metrics output from this analysis. So I would like to use ICA AROMA to help handle motion-related artifacts rather than scrubbing.

I was wondering if you could provide some guidance on how best to incorporate ICA AROMA into my analysis pipeline. 


1. Should I feed in the ICA denoised ICA aroma to perform the additional WM and CSF regression followed by temporal filtering. Although doing multiple sequential regressions seems like a bad idea and might reintroduce spurious noise.

2.. Alternatively, it seems like it would be better to do  ICA AROMA noise (where I feed in the noise components it identified), WM, and CSF regression in one step and then temporal filtering (ICA AROMA should be done prior to any temporal filtering of the data).  Is is possible to do this with CONN? How might I do this?

3. I have not yet seen aCOMP Corr and ICA AROMA combined. In the original ICA Aroma paper, they regressed out mean CSF and white matter signals. I wonder if it is better to do that then do aComp Fo and which would then involve  two PCA/ICA denoting approaches to the data. Was just wondering if you had any thoughts on this.

I appreciate your guidance and expertise!

Emily