[Vv] Using vv/CLITK and elastix to compute target registration error

[Simon Rit]

Hi,
I agree, since you use elastix for registration, it seems that a good
way to do it. As far as I know, vv has its own format for landmarks
but I believe you could quite easily modify the code to allow reading
of VTK points in the VV landmark tool.
Good luck!
Simon

On Wed, Mar 20, 2013 at 7:52 PM, S. Klein <s.klein at erasmusmc.nl> wrote:
> Hi Bilal,
> I would do it like this:
> 1) ask your radiologists to click corresponding points in the ORIGINAL fixed
> and moving images. This gives you two files: fixedpoints.txt and
> movingpoints.txt.
> 2) Run elastix: elastix -f fixed -m moving -p parameterfile -out outdir
> 3) Run transformix to transform the FIXED points: transformix -def
> fixedpoints.txt -tp outdir/TransformParameters.0.txt -out outdir
> 4) now compare the transformed fixed points (which are stored in
> outdir/outputpoints.txt) to the original moving points.
>
> Instead of using the custom elastix input and output point formats (see
> elastix manual, section on transformix), you could also use .vtk (in legacy
> asci vtk format).
>
> Since the transformation is defined as a mapping from the fixed image domain
> to the moving image domain, you can only directly transform points from
> fixed to moving image.
>
> Kind regards,
> Stefan
>
>
> On 20/03/2013 19:06, Bilal Tahir wrote:
>
> Dear VV community,
>
> We are planning to get a radiologist to define corresponding landmarks on
> fixed/moving inspiration/expiration CT lung images using VV. We use elastix
> for image registration. Once the registration is performed, is it possible
> to use VV/CLITK to compute the target registration error (TRE) between the
> fixed image landmarks and the registered moving image landmarks?
>
> Furthermore, we would like the landmarks to be defined on the moving image
> prior to registration so that we can try different parameters to assess
> which gives the best TRE. Perhaps this is outside the scope of the mailbase,
> but has anyone done something similar using Elastix, where the coordinates
> of the expert landmarks are defined prior to registration using VV and new
> coordinates computed depending on the registration algorithm?
>
> Best wishes,
>
> Bilal Tahir,
> James Morrison Researcher in Radiotherapy Imaging
>
>
> _______________________________________________
> vv mailing list
> vv at creatis.insa-lyon.fr
> http://www.creatis.insa-lyon.fr/mailman/listinfo/vv
>
>
> --
> Stefan Klein
> +31 10 7043049
> http://www.bigr.nl/people/StefanKlein
>
>
> _______________________________________________
> vv mailing list
> vv at creatis.insa-lyon.fr
> http://www.creatis.insa-lyon.fr/mailman/listinfo/vv
>