[Insight-users] Level-set methods for image segmentation

[Zein Salah]

Dear friends,
 I am trying to write an image segmentation program based on the level set
classes of itk. I was trying with the ShapeDetectionLevelSetFilter and with
a volume image of  181*217*181. That program had always abnormally
terminated when I call Update with the ShapeDetectionLevelSetFilter-object.
I have just tried with a 64*64*64 image. The program did not stop
abnormally. BUT the result is not as I expect.
 As I understood from the documentation of the ShapeDetection filter

http://www.itk.org/Doxygen/html/classitk_1_1ShapeDetectionLevelSetFilter.htm
 , the inputs are an initial level set and an edge potential image. I think
 the edge potential image created by my program is correct. I looks very
much
 like an edge map of the original image.
 However, I think the initial level set is not correct. it is just an image
 in which a central point (a seed) is black and the other pixel go gradually
 to white as we go further from the seed point. I realy do not know if this
 is correct.
 How should the initial level set look like??

 With many thanks.....

 Zein