[Insight-users] Advice needed in using level-set methods for liver segmentation

[Poh Choon Chien]

Hi all,
 
I have been trying to implement level set methods for liver segmentation from CT images, in particular using the Threshold LevelSets and GeodesicActiveContours:
 
However i have difficulties in selecting the optimal values to use as the scaling parameters for propagation, curvature and advection.... thus i hope to get some advice on what values may be suitable for a liver segmentation problem? And is there a good rule of thumb to position the seed for the fast-marching algorithm to produce the initial contour so as to achieve optimal performance?
 
I will greatly appreciate your help here..  Thanks
 
Leon Poh