FW: [Insight-users] GeodesicActiveContourImageFilter...

[Jean-Philippe Guyon]

Hello Thomas,

I have not been able yet to get a useful segmentation out of the
GeodesicActiveContourImageFilter. It seems that the front that propagates
always goes beyond the boundaries of the object, even though the edge
potential image has quite noticeable edges with values close to zero, and
the surrounding environment has values close to one ( area where the
gradient value is low, refer to the documentation
http://www.itk.org/Doxygen/html/classitk_1_1GeodesicActiveContourImageFilter
.html ). There are still a few discontinuities in the edges through which
the front could eventually propagate, but I was expecting that setting the
smoothness constraint to a pretty high value should prevent the propagation
through these gaps. Since the edge potential image that I feed into the
filter seems to be valid, I assumed that something must be wrong with the
derivative images that I feed into the filter.

Please let me know if you have any idea.

Thank you,

ps:
the parameters I use are the following:
Number of iterations: 50
TimeStepSize: 0.5
LengthPenaltyStrength: 0.5
InflationStrength: 0.1
NarrowBanding: true


-----------------------
Jean-Philippe Guyon
home: (919) 929-4132
work: (919) 843-4921           Department of Radiology, CB# 7515
piloo@unc.edu                  UNC-Chapel Hill
http://www.pilooarena.fr.st    Chapel Hill, NC 27599-7515

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

-----Original Message-----
From: Th. Boettger [mailto:t.boettger@dkfz-heidelberg.de]
Sent: Friday, November 22, 2002 4:21 AM
To: Jean-Philippe Guyon
Subject: Re: [Insight-users] GeodesicActiveContourImageFilter...


Hallo,

If you are already using the GeodesicActiveContour, do you use it with
3D data and did you have any success? I tried the filter, but could not
get any useful results out of it.

Thanks
Thomas Boettger




Jean-Philippe Guyon wrote:
> Hello,
>
> I am using the GeodesicActiveContourImageFilter to segment uterus fibroids
> in 3D. I was wondering if the derivative images that have to be set as
input
> into the filter need to be normalized. It would be helpful if it was
> mentioned in the documentation.
>
> Thank you in advance for your help.
>
> Cheers,
>
> -----------------------
> Jean-Philippe Guyon
> home: (919) 929-4132
> work: (919) 843-4921           Department of Radiology, CB# 7515
> piloo@unc.edu                  UNC-Chapel Hill
> http://www.pilooarena.fr.st    Chapel Hill, NC 27599-7515
>
> -----------------------
>
> _______________________________________________
> 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