Ryan,

The concept of deformable models is a tricky one. I myself have been
dealing with them for a few years still not sure sometimes how thing
should work.

The simplest way to look at this is as follows,

deformable model are governed by external and internal forces. In the
case of simplex meshes, 

internal forces are calculated from mesh vertices and are responsible
for keeping the vertices at a certain distance from each other. i.e,
smoothing the mesh.

The external forces are the tricky ones. Here you calculate some kind of
an edge map of the 3d image, gradient for one and try to move the
vertices of the mesh towards the edges.

so the parameters alpha and beta are very important to set. if you
choose them to large, the mesh will explode and go past beyond edges, if
you choose them too small, the mesh will not do much.

so try saving the gradient magnitude of your image and look at it, you
can also save the gradient image itself and view it in paraview.

In my experience, you must read the papers and experiment with the
parameters over and over again to get them right.

be prepared to be patient!

HTH

Leila

On Thu, 2008-17-07 at 15:45 -0700, Ryan Barr wrote:
> Hi Leila,
> 
> thank you for the prompt reply and answer.
> 
> Unfortunately  I'm clearly still having troubles, even after reading
> at least half a dozen publications by Delingette, I am still trying to
> get my head around this stuff.
> 
> I suppose my greatest confusion at the moment is that in the
> publication you referred me to, they are deforming a simplex mesh to
> conform to an isosurface. However, the DeformableModelSimplexMesh
> program is supposed to load image data (CT, US, MR) from an mdh (+
> raw) file.
> 
> Is there any chance you could provide a quick walk through of the
> program, including a link to the data that is used in File->Load File?
> I've gone through the help file with the program but to no avail (see
> the next paragraph for examples of why not).
> 
> Based on the image you sent me, you loaded a mdh file, set a seed on
> the image and selected Create Mesh. What happens when you Deform Mesh?
> For me either nothing happens at all, or the mesh "explodes". I've
> tried using mdh data from http://public.kitware.com/pub/itk/Data/ but
> with no success. For example, loading Patient01 from
> http://public.kitware.com/pub/itk/Data/LiverTumor/ causes an exception
> (appears there is missing gradient information), or the program
> crashes during PreProcessing with Patient06 ("Failed to allocate
> memory for image." message). On the other hand I've tried sphere.mdh
> and gradient.mdh from
> http://public.kitware.com/pub/itk/Data/Synthetic/ and regardless of
> what settings I use the mesh never changes shape, except in some cases
> where it "explodes".
> 
> Also, are there any other ITK/VTK programs that you might recommend to
> help with mapping a simplex mesh to a vtk (or isosurface) model?
> 
> Thanks again for your time and help,
> 
> Ryan
> 
> > Date: Thu, 17 Jul 2008 12:58:30 -0400
> > From: baghdadi@phenogenomics.ca
> > To: ryan_barr_eac@hotmail.com
> > CC: insight-users@itk.org
> > Subject: Re: [Insight-users] TriangleMeshToSimplexMeshFilter
> > 
> > Hi Ryan,
> > 
> > I read your e-mail twice and am still not sure what your question
> is.
> > 
> > I extensively use all those filter for my project (mouse embryo
> > segmentation) and am quiet sure they work correctly (although there
> > might a bug here or there).
> > 
> > The result of the TriangleMeshToSimplexMeshFilter is what it says it
> is.
> > 
> > It takes triangles and converts them to simplex cells. If you are
> not
> > sure how that works try reading paper
> > http://citeseer.ist.psu.edu/147387.html
> > 
> > and yes all the work on these filters is on the basis of herve
> > delingette's work.
> > 
> > here is a snap shot of what you should see (result of
> > TriangleMeshToSimplexMeshFilter)
> > 
> > HTH
> > 
> > Leila
> > 
> > 
> > On Thu, 2008-17-07 at 09:09 -0700, Ryan Barr wrote:
> > > Hi, I am trying to generate a Simplex Mesh surface based on a vtk
> > > polydata file input.
> > > 
> > > Using the DeformableModelSimplexMesh example as a basis only, I
> can
> > > convert a vtk model (with Polygon cell types) into an itk triangle
> > > mesh and then produce an itk::SimplexMesh using
> > > TriangleMeshToSimplexMeshFilter. I then successfully return it
> back to
> > > a vtk model.
> > > 
> > > However, when I use the TriangleMeshToSimplexMeshFilter it
> produces
> > > something that looks little like the images in Delingette's
> > > publications (e.g. triangles instead of Delingette's hexagons),
> and
> > > moreover, it makes the surface more complex, roughly doubling the
> > > number points and cells, which is once again contrary to what I
> > > thought a simplex mesh should do.
> > > 
> > > Furthermore, nearly all the readings I have seen on simplex
> meshes,
> > > and the origin of the DeformableModelSimplexMesh example itself,
> are
> > > from Delingette.
> > > 
> > > My question is, what is the TriangleMeshToSimplexMeshFilter based
> on,
> > > if not Delingette's work?
> > > 
> > > Also, there are a number of other SimplexMesh filters written by
> the
> > > same author as TriangleMeshToSimplexMeshFilter (e.g.
> > > SimplexMeshGeometry). Are any of these other ones capable of
> producing
> > > something more akin to what Delingette uses?
> > > 
> > > Below are a couple of the Herve Delingette publications I am
> referring
> > > to...
> > > 
> > > Soft Tissue Modeling for Surgery Simulation, 2004
> > > Simplex Meshes: a General Representation for 3D Shape
> Reconstruction,
> > > 1994
> > > 
> > > Thank you,
> > > 
> > > Ryan Barr
> > > Computer Engineering
> > > University of British Columbia
> > > 
> > >
> ______________________________________________________________________
> > > Have fun while connecting on Messenger! Click here to learn more.
> > > _______________________________________________
> > > Insight-users mailing list
> > > Insight-users@itk.org
> > > http://www.itk.org/mailman/listinfo/insight-users
> 
> 
> ______________________________________________________________________