Hi Sarah,

3dFourier only removes linear trends. Consider
3dBandpass instead, which removes quadratic trends.
On top of that, 3dBandpass can remove the other
regressors of no interest. With 3dFourier, you
would have to run it on those regressors separately.

Actually, in afni_proc.py, neither program is used
for bandpassing, as they would not allow for proper
censoring. In afni_proc.py, censoring, bandpassing
and removal of undesirable signals is done in a
single linear regression step. The polort used in
afni_proc.py is at least 1, but is still based on
the length of data.

I should add that using polort 2 should be sufficient
when bandpassing out the lower frequencies.

Note that afni_restproc.py was written by Rayus
Kulicki in Tulsa, and support must come from there.
It is different from (but based on) afni_proc.py.
I will offer replies in terms of afni_proc.py, of
course.

- rick



Edited 1 time(s). Last edit at 09/22/2014 03:09PM by rick reynolds.