[Insight-users] Re: question about WriteDisplacementField in FEM deformableRegistration1

Luis Ibanez luis.ibanez@kitware.com
Wed May 12 23:52:51 EDT 2004


Hi Tina,


A) The FEMRegistrationFilter do not apply any scaling
    to the field components when they are writen using
    the method

               WriteDisplacementField( int )

    The components values are exactly the same ones that
    you obtain by calling

          WriteDisplacementFieldMultiComponent()

    The scalar component are simply extracted from the
    vector image by using the filter

           VectorIndexSelectionCastImageFilter<>


    You will find this code in


        Insight/Code/Algorithtms/
          itkFEMRegistrationFilter.txx

                                lines 582-590




B) The m_FloatImage is returning the an image where every
    pixel represents the Determinant of the Jacobian Matrix
    of the transformation at that location.

In an incompressible medium this value should be 1.0 in all
the pixels. It will be < 1.0 if the transform is expanding,
it will be > 1.0 if the transform is compressing. You never
want it to be negative since it will indicate a folding in
space.

It is a good idea to take a look at this image once you
collect your registration results. In most anatomical cases
you don't want compression nor expansions. Except probably
when you are registering images before and after a graft
has been implanted, or before and after a tumor has been
removed.





  Regards,



     Luis


-------------------
Tina Wang wrote:

> Luis,
>  
> After the registration is done, we could write the displacement vector 
> field in x, y, and z directions, respectively. For example, in writting 
> the displacement x field, do you just extract the x-field from vector 
> image (m_Field) or you have done some scaling based on the displacement 
> in x field?  I also noticed that before warping the image, the 
> m_FloatImage is computed in Function of ComputeJacobian(). But this data 
> is never referenced. What is this data for?
>  
> Thank you very much for your helps.
>  
> Tina
>  
>  
> 
>  





More information about the Insight-users mailing list