[Insight-users] FastSymmetricForcesDemonsRegistrationFunction

[Martin Urschler]

hello all

i came across a new fast symmetric forces demons registration function 
contributed by torsten rohlfing (thanks for that...)

To compare the new fast symmetric forces demons registration function 
with the older symmetric function I made a copy of the 
SymmetricForcesDemonsFilter and replaced the finite difference function 
with the new faster one, calling it FastSymmetricForcesDemonsFilter.
I'm sure that I didn't make a mistake here, a diff of the files proves 
that I only replaced the finite difference functions.

But, my results of matching inhaled and exhaled lung images are very 
different. While the old symmetric demons function gives quite good 
overall results (except vessel structures being destroyed, but that's a 
different topic) the fast symmetric demons function gives results 
showing far off displacements. Further, the metric values are increasing 
most of the time especially for downsampled volumes during the multi 
resolution iterations.

So my question is, has anyone already tested this combination and found 
similar results? Or am I doing something wrong? Is there a bug in the 
fast symmetric forces demons function?
Torsten, could you perhaps give it a try and use your fast symmetric 
forces function with the symmetric demons filter?

thanks,
Martin