[Insight-users] itk::FEM Framework - Adaptive Meshes

[Suresh Joshi]

Hey Vincent,

Thanks! I'll take a look at IA-FEMesh... It seems to be what I'm
looking for. And at the moment it's 2D registration, but I will be
moving to 3D soon enough. I currently work with the built-in mesh with
adequate results, but as my educational background has a lot of
mechanical engineering, I'm pretty familiar with FEM and I just don't
think it would be fair to try to compare between registration methods
and not try an adaptive mesh with the FEM technique.

Interestingly enough, it was your previous post about loading and
converting a mesh which kind of spurred my idea of really trying to
find a mesh generator, after seeing that people have, in fact, done it
before.

I have done a lot of searching for FEM parameters with few results.
Unfortunately, it is not exactly the subject of my research at the
moment, but if by chance I stumble upon interesting results in the
course of optimizing the FEM, I will of course share it with the
world. Offhandedly, my current approach is to use an
evolutionary/genetic algorithm to find optimal parameters for me, just
because I can't spare the time to do an in-depth study on them right
now.

Cheers!
-Suresh

On Mon, Apr 19, 2010 at 2:13 AM, Vincent Magnotta
<vincent-magnotta at uiowa.edu> wrote:
> Suresh,
>
> First, I am assuming that you are trying to perform a 2D registration based
> on the types of elements that you reference. I believe that the only example
> for the ITK::fem registration is in DeformableRegistration1.cxx (2D) and
> DeformableRegistration11.cxx (3D). The easiest way to work with the fem
> registration framework is to have the algorithm create the mesh for you. The
> framework supports a hierarchical framework, but the only contain one
> material property. You will need to create your own mesh to support multiple
> material properties and meshes that conform to anatomical boundaries.
>
> If you can create a mesh and save it in VTK format, then I have previously
> posted information on the insight-users list showing how to load the mesh
> and convert it into a fem::mesh. In 3D I would suggest IA-FEMesh (
> http://www.ccad.uiowa.edu/mimx/IA-FEMesh/). In 2D, I don't have a
> recommendation. The ITK::fem framework is something that needs to be updated
> and hopefully this will be done with the ITK4.0 initiative. This would
> greatly simplify the incorporation of user defined meshes for registration.
>
> I don't know if anyone has done an extensive parameter exploration for the
> fem registration. This would be a great addition to the field and would make
> an outstanding Insight Journal article.
>
> Vince
>
> On 4/18/10 6:01 PM, "Suresh Joshi" <Suresh.Joshi at insa-lyon.fr> wrote:
>
>> Hello all,
>>
>> I'm currently performing registration tests on brain MRI using the ITK
>> deformable registration methods. All is going well, except in
>> FEM-land.
>>
>> Mostly, I know the default mesh created is a rectangular quadrilateral
>> mesh, but after reading an ITK registration tutorial, I saw
>> comparisons between sparse/dense triangular meshes, adaptive
>> quadrilateral meshes and multi-material meshes... Could someone please
>> tell me how these adaptive meshes were obtained/created?
>>
>> Or perhaps, "Anatomy-specific meshes can be created using public
>> domain mesh generators" --> Are there any examples or links to said
>> mesh generators? And if there are also any details on how these meshes
>> are actually converted into the FEM mesh, could someone please let me
>> know?
>>
>> The final question is about the parameters. While there seem to be a
>> lot of parameters to tune, the suggestions only discuss keeping the
>> ratio between rho/E close to 1 and to modify the timestep. And the
>> rest of the parameters? They can be left at default?
>>
>> The reason I wonder is because I'm trying to do an as-close-to-optimal
>> transform parameter comparison, and thus, my laziness (in parameter
>> tuning) shouldn't keep FEM from getting good results  :)
>>
>> Thanks a lot,
>> -Suresh
>> _____________________________________
>> 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
>
> ----------------------
> Associate Professor
> Department of Radiology
> 0453-D JCP
> 200 Hawkins Drive
> Iowa City, IA 52242
> E-mail: vincent-magnotta at uiowa.edu
> Phone: 319-356-8255 Fax: 319-353-6275
> Website: http://www.radiology.uiowa.edu
>
>
>
> _____________________________________
> 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
>