I have a follow-up question for this thread.  I'm currently assessing whether to switch from SPM8 to FSFAST.  As part of this process, I did a comparison between the SPM8 and the preproc-sess motion correction outputs but found they were quite different for the example data I looked at.  Furthermore, I found that the fmcpr.mcdat output and the mcprextreg output are quite different even though they are both generated by fsfast.  The following three charts are for mcprextreg, fmcpr.mcdat, and SPM8 respectively.  Finally, I'm including a correlation matrix (I don't see any correspondences).

The data are actually two separate sessions with a rest break, resulting in a discontinuity at the midpoint.  The SPM realignment was run separately on each and as usual the first volume was the reference volume while FSFAST was run on both sessions as a merged set and I think used the middle volume as its reference volume.  This shouldn't have resulted in such a large difference between SPM8 and FSFAST though (Ardekani et al 2001 reported they didn't differ that much in accuracy for SPM99 and AFNI98 versions and my understanding is that FSFAST uses AFNI's routine) and certainly not between the two FSFAST outputs.

On the other hand, the magnitude of movement in this example is small so I guess my main concern is the discrepancy between mcprextreg and fmcpr.mcdat.  I want to make sure I know what I'm looking at and therefore how to use it properly.  I looked up the documentation on AFNI's 3dvolreg (http://afni.nimh.nih.gov/pub/dist/doc/program_help/3dvolreg.html).  Would I be correct in thinking that mcprextreg results from the -1Dfile option and fmcpr.mcdat results from the -dfile option?  I can take this to the AFNI message board once I'm sure where these files came from.

Respectfully,

Joe