[Insight-users] Deformable Registration Help

[Martin Urschler]

hi

is there a special reason why you insist on fem based registration?

there are other techniques in itk:
- demons based registration in case you can assume brightness constancy
- MI based bspline registration in case your image intensities differ 
between images

both techniques lead to good results if you implement them in a 
multi-resolution manner, and they are more straightforward to use than 
the FEM framework

on your 2d images they should also be reasonably fast!

greetings,
Martin

Michelangelo Paci wrote:
> http://us.f13.yahoofs.com/bc/4415f191ma8e3b5a7/bc/Differenza_t01_tagliato.vtk?bfvBdaEBuLx7_N1w
> 
> http://us.f13.yahoofs.com/bc/4415f191ma8e3b5a7/bc/Differenza_t02_tagliato.vtk?bfvBdaEBgUvjEU2v
> 
> 
> Hi,
> I' ve a doubt about Deformable Registration.
> It's 2 weeks I'm trying to tune the parameters of fem-based deformable 
> registration in order to match the image Differenza_t02.vtk to 
> Differenza_t01.vtk but it's totally usefull.
> So I'm asking if it's possible to obtain a good matching between these 2 
> images, and so a good vector field which leads me to image _t01 from 
> _t02, or if I'm totally wrong in using this tecnique.
> I'm asking you: can I use the DeformableRegistration1 in order to 
> transport the cells in image 02 in the positions they have in image 01?
> 
> Please help me, I could have wasted about 100 hours of work.
> 
> I've appended the URL's of the two volumes I've to match: these are 2 
> successive frames obtained with confocal microscopy.
>