Dear all,
I have some datasets acquired with a *very* oblique prescription (and with only a few slices) and I have some theoretical questions for which I could use some better informed judgment:
1) It seems to me that the motion correction algorithm would perform better when applied *before* any de-obliquing, right?
2) Similarly, if one plans to use the estimated motion parameters as confounds in a linear model (3dDeconvolve), and assuming that motion correction has been performed on the original images (non de-obliqued), it seems to me that the linear model would be better estimated on the images *before* a spatial registration to a template (MNI or Talairach). Unless the amount of variance in the observed data explained by the set of motion parameters is somehow invariant under affine transformation of the data...
any thoughts?
thanks
g.