Hi Bruce,

Thanks for your reply. I have just one more question framed in two different ways for clarifying the situation further:

Does this mean if a certain labeling was applied to the white matter surface, it could produce errors in labeling on the pial surface if the labels were transfered 1:1? Alternatively, does this non-parallelism distort vertex to "region" correspondence (especially at the boundaries of "regions"), i.e. a vertex on the white matter surface belongs to region A but the corresponding vertex on the pial surface might belong to a different region B?

[Assuming "regions" are defined by normals cutting through the cortical sheet at the fundus of sulci or at the crown of the gyri].

Thanks,

Satra

On Mon, 31 May 2004, Bruce Fischl wrote:

Hi Satra,

sorry, there is no such easy relationship as the surfaces are not necessarily parallel. The vertex normals are the ccw cross product of the triangle legs, I believe (you should be able to load the ?h.sphere surface into matlab and check easily enough).

Bruce

On Mon, 31 May 2004, Satrajit Ghosh wrote:

...

Is there an easily quantifiable relationship between the white and pial surfaces? Such as:

pial_vertex = white_vertex + vertex_thickness*vertex_normal;

I understand that vertex_normals could be computed in different ways (Eg, Desbrun et al 2000). If so, could we find out which computation scheme is used (a reference would be sufficient)?

Thanks,

Satra