[Insight-users] Re: ask for help on DeformableMesh3DFilter

[Yan Pingkun]

Hi, Sam,

I experienced the same problem on this filter. I also noticed that you =
would get the same result no matter you set the iteration as 1 or 100. I =
found the problem is caused by the uninitialized stiffness matrix m_K[] =
in the source code. I discussed this problem with the author Ting Chen. =
Actually this is not a bug. The stiffness matrix should be decided by =
the user and there's no default value. You need to call =
"SetStiffnessMatrix" to set it. Otherwise, the filter can't work =
properly.

Unfortunately, the current two examples didn't set the matrix. Hence, if =
you follow them, you can only get what is very similar to your =
initialization.

HTH,
Pingkun

-----Original Message-----
From: insight-users-admin at itk.org [mailto:insight-users-admin at itk.org] =
On Behalf Of Samuel Rodr=EDguez Bescos
Sent: Monday, April 12, 2004 5:20 PM
To: Luis Ibanez; Deliang Ye
Cc: Insight-users at itk.org
Subject: [Insight-users] Re: ask for help on DeformableMesh3DFilter


Hi, Deliang,

I agree with Luis. In adition, I noticed that the surface didn't have
important changes in its evolution when I used   the =
DeformableMesh3DFilter
in an example similiar to yours (to refine a surface model got from a =
region growing  method) . I didn't try any level sets method but I think =
that they could be better option.

With reference to the question about de values of the parameters of the =
filter I think that there aren't a fixed rules, unfortunately we have to =
test and check.

Regards,

Sam

=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
Samuel Rodr=EDguez Besc=F3s
Bioengineering and Telemedicine Department
ETSI Telecomunicacion

Universidad Polit=E9cnica de Madrid
Ciudad Universitaria s/n - 28040 Madrid - Spain

Ph: +34 91 550 04 35 (ext.32) // +34 91 549 57 00 (ext. 332)
Fax: +34 91 336 68 38

Email: srodrigu at gbt.tfo.upm.es
Web: http://www.gbt.tfo.upm.es =
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D


----- Original Message -----=20
From: "Luis Ibanez" <luis.ibanez at kitware.com>
To: "Deliang Ye" <dye at jhu.edu>
Cc: <srodrigu at gbt.tfo.upm.es>; <Insight-users at itk.org>
Sent: Wednesday, April 07, 2004 8:40 PM
Subject: Re: ask for help on DeformableMesh3DFilter


>
> Hi Deliang,
>
> The DeformableModel filter is quite unstable
> and may not be the best options for refining
> an existing segmentation.
>
> You may want to look at the LevelSet methods
> since they provide better mechanisms for
> enforcing estability.
>
> Please take a look at the LevelSet chapter
> in the software guide.
>
>   http://www.itk.org/ItkSoftwareGuide.pdf
>
> Section 9.3, pdf-page 363.
> In particular you may want to look at
>
>   - GeodesicActiveContours
>   - ThresholdLevelSet
>   - CannyEdgesLevelSet
>
> Associated examples are available under
>
>    Insight/Examples/Segmentation
>
>
>
>
> Regards,
>
>
>    Luis
>
>
> ----------------
> Deliang Ye wrote:
>
> > Hi, Luis and Sam,
> >
> > I got a rough segmentation of rat spinal cord by region growing, I=20
> > would like to use deformable mesh filter to refine the result. But=20
> > the program failed, and told me memory access violation. I have not=20
> > changed any parameters in DeformableModel1.cxx  in ITK, what is the=20
> > rule of thumb to
set
> > the parameters, such as stiffness, scale, gradient magnitude, time=20
> > step,
and
> > step threshold in DeformableMesh3DFilter?
> >
> > Please check that attached file,
> >
> > Appreciate you for help!
> >
> > --Deliang Ye
> >
> >
> > --------------------------------------------------------------------
> > ----
> >
>
>
>