[Insight-users] itk LevelSetsv4

[Arnaud Gelas]

Hi Cagatay,

First, I'd like to say that it makes total sense for you to make use of 
the new framework.
One of the main goals of the new framework is to ease the combination of 
any kind of "terms", which is the case here.

In your case, if I understand correctly, you'd need to combine:

  * Chan and Vese internal term
  * Chan and Vese external term
  * Propagation term where you'd provide your own feature image
  * and / or Advection term where you'd provide your own vector field
  * Curvature term (optional for regularization)

The two first terms and the curvature terms are present and validated 
within the new framework.
The propagation term is present in the framework, but the validation is 
coming along and should definitively be part of the upcoming release in 
December.
The advection term is in its way in the framework. As Matt mentioned, 
the patch is in review and still need some validation. It will also be 
part of the upcoming release in December.

While waiting for the upcoming release (which will happen within few 
weeks), I invite you to combine Chan and Vese (internal and external 
terms) with the propagation term.
It may work fine at this point. But your feedback would definitively 
help us to improve the framework and the toolkit !!

To combine the terms, you can have a look at the following tests:

  * 1 level set with Chan and Vese:
      o With a dense representation:
        ITK/Modules/Segmentation/LevelSetsv4/test/itkSingleLevelSetDenseImage2DTest.cxx
      o With a sparse representation
          + Whitaker's formulation: 
            ITK/Modules/Segmentation/LevelSetsv4/test/itkSingleLevelSetWhitakerImage2DTest.cxx
          + Shi's formulation:
            ITK/Modules/Segmentation/LevelSetsv4/test/itkSingleLevelSetShiImage2DTest.cxx
          + Malcolm's formulation:
            ITK/Modules/Segmentation/LevelSetsv4/test/itkSingleLevelSetMalcolmImage2DTest.cxx
  * 1 level set with Propagation term:
      o ITK/Modules/Segmentation/LevelSetsv4/test/itkSingleLevelSetWhitakerImage2DWithPropagationTest.cxx


In the case if you want to see the evolution of your level set through 
the iterations, you may also want to have a look at the tests in 
ITKLevelSetsv4Visualization.

If you have questions, or remarks, please let us know.

Thanks,
Arnaud


On 12/09/2011 02:06 AM, cagatay bilgin wrote:
> Thank you very much for the clarifications
> and the pointers Matt. I checked out the
> latest release a couple of hours ago and
> see a lot of changes as you pointed out.
>
> There is a GeodesicActiveContour class in
> the (old) LevelSet directory, which makes me
> believe that it wouldn't be too hard to come up
> with the corresponding term class for the new
> LevelSet framework. It is still not clear to me
> how to accomplish this though, I am completely
> puzzled with how it was done before.
>
> Cagatay
>
> On Thu, Dec 8, 2011 at 6:49 PM, Matt McCormick 
> <matt.mccormick at kitware.com <mailto:matt.mccormick at kitware.com>> wrote:
>
>     Hi Cagatay,
>
>     On Thu, Dec 8, 2011 at 8:55 PM, cagatay bilgin <bilgincc at gmail.com
>     <mailto:bilgincc at gmail.com>> wrote:
>     > I am trying to understand the new level set framework
>     > and willing to use it in our project. Being able to add
>     > terms to the level set function would be really helpful
>     > I think. Is this the direction that ITK will be taking ?
>
>     Yes, the new level sets frame takes that direction, and it is a
>     flexible and powerful approach that allows for new avenues of
>     research.
>
>     > I see two directories LevelSets and LevelSetsv4, one
>     > having term classes and the other not having them
>     > which makes me wonder how the framework will be
>     > working. It is a little hard to find directions and documentation
>     > for me as the framework is still being developed.
>
>     Currently, the old levelsets framework is in the ITKLevelSets module,
>     and the new framework is in the ITKLevelSetsv4 module.  They are not
>     compatible, and the term-based framework is ITKLevelSetsv4.
>
>     Suggestions for disambiguation of the old and new framework are
>     welcome.
>
>     >
>     > I am trying to build an application that implements the
>     > geodesic chan vese algorithm which uses the internal
>     > and external terms from the chan vese algorithm and
>     > would also include the gradient information from the
>     > geodesic active contours to snap to the edges better.
>
>     The Chan and Vese internal and external terms have been implemented.
>     An advection term is in the works:
>
>     http://review.source.kitware.com/#change,2914
>
>     Arnaud may have better suggstion on how to proceed with the
>     implementation you desire.
>
>     You may find the MICCAI tutorial useful.
>
>     http://midas.kitware.com/community/view/65
>
>     There is a copy of the presentation, but also a VirtualBox virtual
>     machine containing the code and examples.
>
>     The code has evolved since MICCAI.  Most notably, the api has been
>     greatly simplified by not requiring different component class names to
>     correspond to each type of level set datatype (dense image, Shi,
>     Whitaker, Malcolm).
>
>     Thanks,
>     Matt
>
>
>     > Should I be working on the filters defined in LevelSets
>     > or in LevelSetsv4? Is there a geodesic term class
>     > defined somewhere that I can use with other terms in
>     > LevelSetsv4 directory ?
>     >
>     > Any help would be appreciated
>     >
>     > Thank you,
>     > Cagatay
>     >
>     > _____________________________________
>     > Powered by www.kitware.com <http://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
>     >
>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/pipermail/insight-users/attachments/20111209/882ad93b/attachment.htm>