I see you have pretty large inhomogeneities in your images. Bias correction
will certainly help here, and maybe you don't need the skull stripping.
Bias correcting before coregistration was actually mentioned on the list not
so long ago...
On Monday 20 October 2008 19:44, Neil Chatterjee wrote:
> Hi SPMers,
> I've noticed that I have more luck with coregistration if I first bias
> correct and skull strip my T1 structural images with the VBM toolbox.
> See attached pictures for two examples (one is very noticeable, the
> other subtler). I've looked through the email archives and the spm5
> manual and haven't seen any mention of it so far.
> Does anyone know why my coregistration might be off in the first place?
> Also, is this a valid workaround that has been tried before?