Hello,

I've been using spmregister in order to coregister a static PET image to an MPRAGE, typically with good results. A representative run is as follows:

spmregister --mov $file.nii --s $subj --reg SUV_lin_T1.dat --fsvol orig

We then convert *.dat to *.mat with tkregister2, and apply the *.mat to the PET with FLIRT.

However, the registration was unsatisfactory for a recent subject. While working through a couple different coregistration options, I tried running coregister (estimate and reslice) in SPM. The result from this was quite good.
I was confused why this would give a different result from spmregister when I'm not changing any of the default parameters in SPM. Any ideas on why running a coregistration directly through SPM would give a different result?

And as a secondary follow-up, are there any tools available to convert an SPM transformation matrix (e.g. x from x=spm_coreg) to a *.mat that can be used in FSL FLIRT?

Thanks,