[Insight-users] Fast symmetric demons registration

[Luis Ibanez]

Hi Ganesh,


1)  The Demons methods in ITK do not use control points.

     The compute a dense deformation field that has the same
     pixel resolution (and image grid) of the fixed image.


2) When dealing with multi-resolution methods you want to
    start fine tunning parameters level by level.  That is, run
    first only the coarsest level, and fine tune its parameters
    until you get a satisfactory result, only then, move to the
    next resolution level.


3) If the image is not deforming enough: reduce sigma
    If the image is deforming too much: increase sigma.


4) Applying histogram matching will be very important if
    your images have different scales of gray levels.

    Have you compared the histograms of the Fixed and
    Moving images ?

    Demons relies a lot on your images having almost
    the same intensity scaling, since it computes optical
     flow at every pixel.


   Regards,


         Luis


-----------------------------------------------
On Wed, Mar 10, 2010 at 4:07 PM, Ganesh Narayanasamy
<nganesh76 at hotmail.com> wrote:
>
> Hi,
>  I am keen on using Vercauteren's Fast symmetric demons registration
> (http://hdl.handle.net/1926/510) . Even though I have worked on image
> registration, I am very new to the demons method, to ITK as well as to C++.
> Can someone guide me with regard to these 2 issues:
> a) When I registered 2 3D-CT images of phantom, I got this response. The
> output image is not that good and I would need to tweak with some of the
> parameters. Can you let me know which ones to tweak.
>
> C:\Ganesh\DiffeomorphicDemons\Build\Debug>DemonsRegistration.exe -f
> S2_SE3.mha -m S4_SE3.mha -o Demon-output-S2SE4fS4SE4m.mha
>
> Starting demons registration with the following arguments:
> Arguments structure:
>   Fixed image file: S2_SE3.mha
>   Moving image file: S4_SE3.mha
>   Input field file:
>   Input transform file:
>   Output image file: Demon-output-S2SE4fS4SE4m.mha
>   Output field file:
>   True field file:
>   Number of multiresolution levels: 3
>   Number of demons iterations: [ 15 10 5 ]
>   Deformation field sigma: 1.5
>   Update field sigma: 0
>   Maximum update step length: 2
>   Update rule: exponentialize and compose (Diffeomorphic)
>   Type of gradient: symmetrized (ESM for diffeomorphic and compositive)
>   Use histogram matching: false
>   Algorithm verbosity (debug level): 0
>
>
> b) Is there a way to select control points in the 2 3D images?
>
> Thanks,
> Ganesh
>
>
>
> _____________________________________
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at
> http://www.kitware.com/opensource/opensource.html
>
> Kitware offers ITK Training Courses, for more information visit:
> http://www.kitware.com/products/protraining.html
>
> Please keep messages on-topic and check the ITK FAQ at:
> http://www.itk.org/Wiki/ITK_FAQ
>
> Follow this link to subscribe/unsubscribe:
> http://www.itk.org/mailman/listinfo/insight-users
>
>