BrainVISA

Hello,

I'm interested in cortical surface parameterization and was using Brainvisa to generate brain meshes for this purpose. The problem is that the resulting meshes are not necessarily genus zero surfaces -- upon closer inspection, there are handles hidden in an otherwise well-behaved looking mesh. Since it is not a topological sphere, we are unable to perform spherical parameterization.

I've read a paper published by the LSIS lab that parameterizes the surface onto a rectangular planar domain, which still would require a certain topology from the surface. I was wondering if you ran into this issue and how you dealt with this issue? Is it an issue that arises from earlier in the pipeline? I have used Morphologist 2011 for bias correction and brain mask segmentation with default settings, which seemed to give reasonable results. I have only provided my own brain masks in addition to T1 images.

Thank you,

--Alex

Hi Alex,

Are you talking about grey/white meshes, or grey/CSF meshes ? In Morphologist 2011, only G/W meshes were supposed to have a spherical topology (even if I sometimes happened to have a few doubts about them).
Morphologist 2012 brings significant improvements in G/W segmentation and in the quality of meshes. Moreover the G/CSF mesh now has the same topological constraints as the G/W one, and should have a spherical topology. However there is no guarantee against intersecting triangles in a mesh, but it is a different problem; and there is now a cleaning pass which removes sharp peaks on the mesh, which reduces (even if it does not remove totally) occurrences of intersecting triangles.
The LSIS surface parameterization also assumes that meshes are spherical.

Denis

Denis,

Yes, I am using the G/W meshes. I re-ran the mesh generation using the 2012 procedures in BrainVISA 4.2.1. The meshes look a lot nicer and the handles that made some meshes non-genus zero went away.

Thanks for the tip,

--Alex

Good news. Just, I'm not completely sure if there is a good reason for this specific kind of improvement... The meshes are nicer, for sure, and thus are less subject to topological defects, but the mesher is the same (maybe used a bit differently).
Denis

In fact the procedure to get the mesh has changed so there is an explanation.
I am surprised there was still a problem in the older version, but maybe it is
something that happened temporarily in the 2011 version because of a bug.