[Insight-users] Mesh and FEM

[Luis Ibanez]

Hi Lara,

Your observation is correct,

The FEM framework do not use the
itkMesh for its internal computation.

There have been attemts to integrate
the itk::Mesh into the FEM structure
but they have not proved a definite
advantage in doing so.

The itk::Mesh is used mainly by the
deformable models and the cellular
agreggates methods for segmentation.

Aljaz and Tessa are the main developers
of the FEM framwork. They could probably
be able to give you some guidance in
its use.


Regards,


    Luis


-----------------------

Lara Vigneron wrote:
> Dear all,
> 
> I have several questions about itkMesh and the FEM framework.
> 
> I saw that the "representation" of a mesh by itkMesh is made with points 
> and cells.
> 
> On the other hand, I want to use the itkFEMRegistrationFilter like in 
> the DeformableRegistration1.cxx example. For the moment, I would like to 
> create a mesh out of ITK and push into via the  CreateMesh method of the 
> itkFEMRegistrationFilter. However, I think the "representation"  of a 
> mesh in the FEM framework is made via arrays representing elements, 
> nodes and materials,  and I didn't find any link with the representation 
> of the mesh with points and cells (itkMesh). Is it correct?
> 
> And in this case, how can I use for instance the BoundaryFeature concept 
> of itkMesh with FEM?
> 
> Also, I saw that to enter a mesh with the CreateMesh method, you have to 
> inject a file characterizing the mesh via the parameter file (it is 
> asked for the MeshFileName near the end of the file) used in the 
> itkFEMRegistrationFilter ( an example of the parameter file is given 
> page 208 of the SoftwareGuide). Is there any example of a MeshFile 
> somewhere to see the synthax to use?
> 
> Lastly, I want to use several elements types with different material 
> properties. Is it sufficient if these different elements are noticed in 
> the MeshFile (see just above) and not in the parameter file where, for 
> instance, only one E value is given?
> 
> Thanks a lot for any help...
> 
> Lara
>  
> 
> -- 
> Lara VIGNERON
> 
> Signal Processing Group
> 
> University of Liege
> Institut Montefiore
> Sart-Tilman, Batiment B28
> B-4000 Liege
> BELGIUM
> 
> Tel: +32-(0)4-366.26.42
> Fax: +32-(0)4-366.29.84
> Lara.Vigneron@ulg.ac.be
> 
>