[ITK] Deformable Registration Field Definition

[Matt McCormick]

Hi James,

The ITK transforms map from the fixed to moving.  Also, it helps to
conceptually to also use the more precise term "displacement" instead
of "deformation", so

  ‘From displacement field’
For each voxel in fixed CT, the vector points to the location in the
moving CT from which we should grab and place here

HTH,
Matt

On Thu, Aug 14, 2014 at 5:14 PM, Armstrong, Hirotatsu
<HIROTATSU.ARMSTRONG at ucdenver.edu> wrote:
> Hello,
>
>
>
> I was wondering if it would be possible to get information of exactly how
> the deformable registration field is defined, specifically for the daemons
> filter.  Assuming we have a fixed CT and a moving CT of the same number of
> pixels in each dimension, the deformable field could be defined as
>
>
>
> ‘To deformation field’
>
> For each voxel in moving CT, the vectors point to the location this voxel
> should be mapped in the fixed CT
>
>
>
> Or
>
>
>
> ‘From deformation field’
>
> For each voxel in fixed CT, the vector points to the location in the moving
> CT from which we should grab and place here
>
>
>
> In addition, each of these could be defined as either +/- depending on how
> you define the starting location of the pixel shifts.  I am generating these
> deformation fields and saving them as VTK files, then applying them in my
> own implementation.  I have switched definitions and tried multiple ways to
> apply the deformation field, and no single method produces a result that is
> obviously correct or obviously wrong.  Any information on how the
> deformation field is actually defined would be appreciated.  Thanks
>
>
>
>
>
> James Armstrong, Ph.D.
>
> PostDoctoral Fellow, Physics
>
> Department of Radiation Oncology
>
> University of Colorado School of Medicine
>
> 1665 Aurora Ct, MS F706
>
> Aurora, CO 80045
>
> 720-848-0257
>
>
>
>
> _______________________________________________
> Community mailing list
> Community at itk.org
> http://public.kitware.com/mailman/listinfo/community
>