[Insight-users] Problems with Level Set Methods

[Christian Rieck]

Hi again

I have some problems regarding the use of Level Set Methods. I have read the
manual and I am now trying to use one of the example files,
LaplacianSegmentationLevelSetImageFilter.cxx.

With the 2D test image, BrainProtonDensitySlice.PNG and
VentricleModel.PNG they use in the manual I get the same results as
they do. However when I try with a
3D volume it does not go to well.
Images and code can be found at http://www.idi.ntnu.no/~rieck/dipp/

speed.PNG is the model I give
the filter, the initial segmentation for it to enhance. in.PNG
 is the image I want to segment. Levelset.PNG is the result of the
LaplacianSegmentation. From code:

diffusion->SetInput( image );
laplacianSegmentation->SetInput( model );
laplacianSegmentation->SetFeatureImage( diffusion->GetOutput() );
thresholder->SetInput( laplacianSegmentation->GetOutput() );

//debug
//ImageType::Pointer debug = laplacianSegmentation->GetOutput();
//debug->Update();
//return debug;

debug is here levelset.PNG.

This is sent trough a thresholding filter to get the final result, seen in
result.PNG.


I have done the same thing with the 2D test image, and the corresponding
laplacianSegmentation->GetOutput (VentricleLevelSet.png) does not look
anything alike levelset.PNG.

I am doing something fundamentaly wrong? I have tried with a variety of
parameters as well, without any luck. The parameters I used for these
example images were: file1 file2 5 1 1 -2 0.


regards,
Christian.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://public.kitware.com/pipermail/insight-users/attachments/20070423/9d519cad/attachment.htm