Page 1 of 1

Topological correction step

Posted: Tue Jan 30, 2018 3:45 pm
by Tordi

I'm working on algorithms of brain segmentation, in particular the cortex. My interest in using AIMS commands is to obtain a mesh of my results, which are usually fuzzy (values between 0 and 1, thus not binary). My inputs are a T2w MRI and its cortex segmentation. I have tried AimsMesh before a format conversion and a thresholding and it works but without topological correction, so I think that it could improve the final mesh.

Could you tell me please how can I add a topological correction step given my inputs ? Which commands allow us to correct it and which extra-parameters do they need ? I tried to reproduce the Morphologist pipeline without success and I don't know in which step it goes wrong.

Thank you.

Re: Topological correction step

Posted: Tue Jan 30, 2018 5:34 pm
by riviere

Well, you might try using VipHomotopic, which is the command we are using for morphologist grey/white interface correction. Its use may be not so straightforward since we normally use it in a somewhat specific context, and I actually don't remember exactly if it has been modified out of a general context (and if yes, to which extent)... As I remember it grows an interface (homotopic dilation or erosion) with fixed sphere topology towards a segmented target object with possible topological errors, so this could be more or less what you need. The "problem" is that it also performs some segmentation, using fixed labels values (100/200) etc.

Or maybe VipHomotopicSnake could be more suitable for you actually...

I don't have a clear answer because we don't have a command which performs only the topological correction step...