[Insight-users] Level set segmentation boundary behavior?

[Sah Rayman]

I was wondering of ITK can treat image boundries as
"infinity" instead of a real edge?

Here is what I meant, in
http://us.f3.yahoofs.com/users/42f3440bz69c8d65/94c5/__sr_/87b5.jpg?ph0i08CBraF5RX7n

Black is the full image. white is my initial.
Propagation term is 1 everywhere, curvature term
weight is 10. And the level set (using
SegmentationLevelSetImageFilter) will converge as the
red curve. 

What's needed in my case is to have level set grow to
the full image. Is that doable in ITK?

Looks to me the problem is how to extrapolate the
level set data outside the image. In Matlab Level Set
Toolbox (http://www.cs.ubc.ca/~mitchell/ToolboxLS/), I
can use @addGhostExtrapolate as the extrapolation
method and it works just fine. 

Can anyone tell me how that could be done in ITK?



__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com