[ITK-users] Largest Connected Region

[Michael Pinkert]

Hi everyone,

I am using the Otsu algorithm as part of a pipeline to segment breast
images, but the results consistently have several holes and isolated regions
of noise, both of which are large problems I am looking to deal with.  I
intend to use morphological operators for the hole problem, but I have not
had luck finding a method to eliminate the isolated regions of noise while
preserving the remainder of the image.  

The breast is a connected region in the results, and the noise is always a
much smaller region.  Is there a filter that could obtain the largest
connected region from from a binary image, so that the breast may be
extracted and the noise removed?

I have looked at some of the various connected region algorithms, but they
require seeds in order to function, and I have not seen one that can
automatically isolate the largest connected region from a binary mask and
make it as an output.

If none of the existing filters work I intend to write a new one to do so,
but I thought it would be best to check first.  

Best regards,
Michael Pinkert



--
View this message in context: http://itk-insight-users.2283740.n2.nabble.com/ITK-users-Largest-Connected-Region-tp7585889.html
Sent from the ITK Insight Users mailing list archive at Nabble.com.