[caret-users] Remove islands?

[Donna Dierker]

On 04/21/2009 04:56 PM, Dav Clark wrote:
> Hi all,
>
> I'm a relatively new Caret user... so I am still learning the basics.   
> Currently, with multiple brains, I am able to run the segmentation,  
> and things look pretty reasonable (except that much of the skull is  
> also identified as white matter).
>   
If you are using SureFit Operations to run the segmentation, then you 
should be getting a single object at the end of this process.  And if 
the skull is still connected, then there is a bridge (usually eye fat) 
connecting the cortex to the skull.
> If I do a topology report, there are a large number of objects.  In  
> the segmentation tutorial, it suggests that you can use "remove  
> islands" to get rid of all but the largest of these.  However, this  
> menu option seems to do nothing at all.
>   
Doing Volume: Segmentation: Remove Islands should, indeed, remove all 
but the largest object in your currently loaded segmentation.

But it will not disconnect skull from cortex, if there is a bridge of 
white/red voxels connecting the two.

And you should not get a segmentation with multiple objects after 
running SureFit segmentation.  There are plenty of intermediate volumes 
that fit this bill, but not the final segmentation.  Several steps 
ensure it is a single object (until you start patching, of course -- 
then all bets are off).
> Am I confused?  Or is something not working right?
>   
Something isn't right.  Did you crop the initial anatomical volume to 
left or right hemisphere?  Caret needs this to be done to segment properly.

If you want to upload your cropped anatomical volume (preferrably in 
NIfTI form), I can try it on my end:

http://pulvinar.wustl.edu/cgi-bin/upload.cgi
> I can replicate this problem on two different computers with two  
> different versions of Caret (current stable, and the funny version  
> that fixes the obscure font bug on linux).
>
> Thanks!
> Dav
> _______________________________________________
> caret-users mailing list
> caret-users at brainvis.wustl.edu
> http://brainvis.wustl.edu/mailman/listinfo/caret-users
>