[Insight-users] No important changes in DeformableMesh3DFilter

Samuel Rodríguez Bescos srodrigu@gbt.tfo.upm.es
Mon, 20 Jan 2003 16:16:12 +0100


Sorry for the previous message. I didn't write anything!!!!!!!!!


Hi Thomas

The Sigma Value I'm using is 3.0. Do you think that This parameters must be
lower?.

Thanks for your help.

Sam
----- Original Message -----
From: "Th. Boettger" <t.boettger@dkfz-heidelberg.de>
To: "Samuel Rodríguez Bescos" <srodrigu@gbt.tfo.upm.es>
Sent: Monday, January 20, 2003 3:54 PM
Subject: Re: [Insight-users] No important changes in DeformableMesh3DFilter


> The parameters seem to be a bit low to me. I tried the filter on liver
> data. My number of steps (step threshold) is at least 100 and the time
> step was between 2 and 10. Which sigma do you use for the gradient
> filtering? The mesh cannot be deformed very much when it isn't near any
> significant gradient. By the way I experienced that the output is
> invariant to changes in the stiffness parameters...
>
> Thomas
>
> Samuel Rodríguez Bescos wrote:
> > Hello everyone.
> >
> > Firstly I would like to congratulate to Luis,Will, Lydia Josh and the
> > Insight Consortium in general  for  make the
> > ITKGuide. It is a very usefull document and It makes everything easier.
> >
> > Now, I have a question about the DeformableMesh3DFilter.
> >
> > I'm using this filter with a initial mesh derived from the Marching
cubes
> > implemented in ITK (BinaryMask3DMeshSource class) of  the binary image
> > resulted from a initial segmentation of the liver using the confidence
> > connected image filter. The gradient image is  the image before applying
the
> >  itkGradientRecursiveGaussianImageFilter to the initial image.
> >
> > The parameters of the Deformable model filter are:
> >
> > Stiffness= [0.00001, 0.04];
> > Step Threshold=1
> > Time Step=0.10
> >
> > The Final is very similar to the initial mesh.
> > Do anybody know what could happen?.
> > Could the others demormable models filters like ballon force have better
> > behaviour?.
> >
> > Thanks in advance.
> >
> > Sam
> >
> >
> > _______________________________________________
> > Insight-users mailing list
> > Insight-users@public.kitware.com
> > http://public.kitware.com/mailman/listinfo/insight-users
>
>
>
> --
> Dipl.-Inform. Thomas Boettger
> Deutsches Krebsforschungszentrum         (German Cancer Research Center)
> Div. Medical and Biological Informatics H0100    Tel: (+49) 6221-42 2328
> Im Neuenheimer Feld 280                          Fax: (+49) 6221-42 2345
> D-69120 Heidelberg                            e-mail: t.boettger@dkfz.de
> Germany                      http://www.dkfz.de/mbi/people/thomasb.shtml
>