Danny Wrote:
-------------------------------------------------------
> I was wondering if there is a "best" point during
> preprocessing to do the registration to a standard
> template? I know some programs do it before
> subject-level statistics, while others do it
> after.
>
> I am currently running everything within
> AFNI_proc.py. I am wondering if I can do all the
> preprocessing (e.g., motion & slice-time
> correction, blurring, convert to mean % signal
> change, etc.) using proc.py, then do registration
> in a separate script (Qwarp or FSL's FNIRT). Then
> run 3dDeconvolve after registration. Will this be
> appropriate?

Assuming you want to register EPI images I would not recommend it; it seems too 'late' in the pipeline. Data converted to percent signal change has little or no anatomical structure. Even after blurring you loose detail which will probably make the registration worse.

For coregistration I would therefore recommend to use the data from after the motion correction step, but before blurring and signal normalization.

*If* your anatomical volume is registered to your EPI already (in some labs the EPI volume nearest in time to the anatomical scan is used as the target for motion correction) then you could also consider using the anatomical volume for template coregistration. This relies however on the assumption that the subject did not move between the EPI and anatomical acquisitions.