Lesion segment inverted/reversed

[Jenny Lee]

Hi itk-SNAP users, 

I wondered if anyone has had a similar problem. 

I've been using itk-snap for automatic segmentation with region competition snakes, following the instructions here. 

It's worked nicely to segment stroke lesions in two participants' T1-weighted MRIs already. 

In more recent scans, the 'segment 3D' process seems to run smoothly for 3/3 steps and the evolution of the bubble is contained well within the lesion. However after clicking 'finish', the segmentation seems to be inverted, with everything but the lesion selected. The problem persists if I try to use manual segmentation tools, too: all area outside a drawn shape will be selected instead of inside it. 

Please see the image attached. Has anyone else encountered this problem or has suggestions for how to resolve it? The only real difference with this lesion is that it's bigger than the last 2 where the process worked well. All MRIs are T1, collected from the same scanner. 

Many thanks in advance for any help. 

Best, 

Jenny 

[Padriac “Paddy” Hooper]

Hi Itk-SNAP users,

I was wondering if anyone has a solution to Jenny's enquiry.

I have trying to manually segment clumps/agglomerates in my phantom samples.
I can only do this using the segmentation tool in ITK-SNAP (since the phantom walls need to be avoided!).

Thank you very much!


Best,
Paddy