general-discussion > n_trends in fmrilm
Showing 1-2 of 2 posts
Display:
Results per page:
Jun 17, 2011  06:06 AM | Crystal Erickson
n_trends in fmrilm
I have a quick question for you. In the n-trends portion of  fmrilm, if we used NIAK what should we put there. The default is [3 1 1] but I remember you at one time saying to use [0 0 1]. Is that still correct? 
Jun 17, 2011  06:06 AM | Felix Carbonell
RE: n_trends in fmrilm
Hi Crystal:
The definition of n_trends in fMRIStat still depends on the user modeling. The default [3 1 1], means cubic (3) polinomion for temporal trends, while the others (1) mean spatial trend and proportional scaling correction. The spatial trend refers to the removal of the global spatial average. For resting state it is still a matter of ongoing debate. I personally do not like such kind of corrections because they introduce too many data manipulations. For task-related studies I would use [3 0 0] or even [1 0 0] assuming that NIAK did a correct physiological noise removal (filtering and CORSICA)
Regards, Felix.