> 1. I'm trying to reduce the amount of smoothing used in the normalisation
> protocol. Is any way to do it? Which is the minimum allowed by the program?
I think this one can just be answered by empirical exploration. My gut
feeling is that it would work best without smoothing if the templatates
consisted of the average of many subjects. If the template is a single
subject, then I think you would need smoothing.
Convergence would be slower without smoothing, and I would expect there to be
more local minima (an ideal solution would be some kind of coarse to fine
schedule).
>
> 2. I would like to know if there is any toolbox/library that would
> decomposed de affine transforation matrix M in rotations, shear and
> translation.
Try
spm_imatrix(M)
Best regards,
-John
|