If the voxel sizes are incorrectly specified, then scaling may help. If the
voxel sizes are set correctly, then we know how to scale the images along
each axis. Estimating it is likely to give solutions slightly different from
these known values, so will be a bit worse than just doing rigid body
registration (unless the motion of whatever organ is in the image is not
actually rigid - the brain can flop about in the skull).
Best regards,
-John
> The coregistration in SPM use rigid body tranformation, 3 rotations and 3
> translations.
> Why doesn't the algorithm perform scaling? To between modality
> coregistration wouldn't the nonrigid tranformation be the better option?
|