The number of basis functions depends on the FOV of the image that is matched
to the tissue probability maps. The number of basis functions actually used
is defined by:
max(round((V(1).dim(1:3).*vx)/opts.warpco),[1 1 1]);
i.e. the field of view (in mm) divided by the user specified cutoff. If
values of [1 1 1] are used, then only translations are modelled.
Best regards,
-John
On Wednesday 24 May 2006 02:08, Hedok Lee wrote:
> Ming,
>
> I tried affine only normalization by setting "non-linear frquency cut
> off"="inf" Alternatively, you can try "non-linear iterations=0", but I
> haven't tied that.
>
> I'm also wondering about the unit of period in "non-linear frquency cut
> off". I appreciate if someone can help me with this. Is this given by
> the number of voxels so that if I were to normalize with t1w template
> (91x109x91), having cut off frequency=25 is 4x5x4 harmonics of cosine
> functions?
>
> thanks,
>
> Hedok
>
> mintnuclear wrote:
> > Dear SPMers,
> >
> > In SPM2 I had the option to choose Affine only during normalisation. In
> > SPM5, a high cutoff (by e.g. 100) value produces few basis functions but
> > I would like to cancel nonlinear deformation at all.
> >
> > Many thanks
> > Ming Chi
|