Gitelman's question about how REST analyzes covariables

Submitted by admin on Wed, 03/11/2009 - 09:11

Dear SONG Xiao-Wei and YAN Chao-Gan:

Thank you for making REST fmri analysis software available. It has
been very helpful in my analyses.
I have a question about how the software analyzes covariables.

I notice that you add both a linear and a mean regressor to the
covariable matrix.
I understand why you add the mean, but why is the linear regressor also added?
Wouldn't this have been taken care of by detrending the data?

thank you,
-----
Darren Gitelman

 

Submitted by Xiang-Yu Long on Wed, 03/11/2009 - 14:39

I think using the  linear regressor is to make sure the time courses and covariables were both detrened, sometimes the covariables may including head motion parameters and these may have some linear trend.

Submitted by admin on Wed, 03/11/2009 - 15:11

I don't think REST would detrend also for covariables by linear regressor.

I think detrend would be optional if there are covariables defined.