[Insight-users] Semiautomatic liver segmentation

[Joshua Cates]

Hello Ivan,

To examine the feasibility of watersheds, you might try the 
SegmentationEditor application in the InsightApplications repository.  It 
is a little work to build because it requires VTK, but will guide you 
through a 3D seg of your data.  You could also test out the 
threshold-based level-set segmentation in the 
ThresholdSegmentationLevelSetFltkGUI, which is a sort of level-set analog  
of a neighborhood connected filter.  The application is 2D but might give 
you some indication of whether that method would work for you.  

Other things you might try: generate a prior using your connected 
component filter and refine using a level-set filter such as the Canny or 
Laplacian, which will fit your surface to edges in the image.

For preprocessing for any of these segmentation filters, I would suggest 
5-10 iterations of an edge-preserving smoothing such as anisotropic 
diffusion or curvature-limited smoothing 
(Gradient/CurvatureAnisotropicDiffusionImageFilter).  Set the time step to 
.1 for gradient, .06 for curvature and start with a conductance of around 
.5.  You will need to adjust the conductance depending on what strength 
edges you need to preserve.

hope this helps,

Josh. 

______________________________
 Josh Cates			
 Scientific Computing and Imaging Institute
 University of Utah
 Email: cates at sci . utah . edu
 Phone: (801) 587-7697
 URL:   http://www . sci . utah . edu/~cates


On Thu, 23 Oct 2003, Ivan Macia wrote:

> Dear ITK users
> 
> Im working on a semiautomatic segmentation of the liver surface using MRI
> images. Im trying to test several segmentation methods using the algorithms
> available in ITK.
> 
> One of the approaches I have taken is using the Neighborhood Connected
> Threshold filter to segment the liver after a first preprocessing of the
> image using Curvature Flow filter. The results are reasonably good but still
> the segmentation result includes many areas outside the liver region. There
> is also the difficulty of selecting the thresholds for the Neighborhood
> filter due to the variability of the intensities with MRI images.
> 
> I am currently working in the implementation of Geodesic Active Contours
> following the pipeline in the ITK guide. I am also thinking in trying
> Watershed filter.
> 
> Has anyone had experience with liver segmentation with MRI? I have seen that
> there is an ITK project going on about this.  I would like to know, if
> possible, which methods or preprocessing steps have given good results or
> any good reference to the topic.
> 
> Thank you very much in advance for your help
> 
> Ivan Macia
> 
> __________________________________
> Ivan Macia
> Ing. Industrial
> Ing. Automatica y Electronica Ind.
> VICOMTech - Visual Interaction and
> Communication Technologies Center
> Paseo Mikeletegi 57
> 20009 San Sebastian
> Spain
> Tel: +34 943 30 92 30
> Fax: +34 943 30 93 93
> e-mail: imacia at vicomtech . es
> http://www . vicomtech . es
> 
> *** member of INI-GraphicsNet ****
> http://www . inigraphics . net
> __________________________________
> 
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk . org
> http://www . itk . org/mailman/listinfo/insight-users
>