[elastix] penalty on registration

M.Staring at lumc.nl M.Staring at lumc.nl
Tue Mar 2 11:28:20 CET 2010

Want to make one point a bit more clear. 

> Actually, it is a bit different, which is due to the B-spline 
> transform.
> When you move a B-spline control point, it not only has an 
> effect at the position of that control point, but also in its 
> neighbourhood, due to the (compact) support of the basis 
> functions. The size of this neighbourhood is related to the 
> order of the B-spline, which is 3 in your case, so the 
> neighbourhood size in which the transformation is different 
> from MI without the penalty term is 4 * gridspacing. I 
> mention that in section II-C of the paper, and in the conclusion.

The size of the neighbourhood is 4 * gridspacing, which means that the
radius is 2 * gridspacing. So, 2 * gridspacing away from your rigid
region you should not notice any effects anymore.


More information about the Elastix mailing list