Hello Freesurfers,
I was wondering if there is information embedded in the annotation/label files that could be used to determine whether two labels represent anatomically adjacent regions? For example, I have parcellated a bunch of participants using connectomemapper into 1000 ROIs. I was just now reading Power et al. (2011), who looked at functional connectivity, but specifically excluded connections between spheroid ROIs that were less than 20mm apart. Analogously, I might wish to exclude connections between adjacent labels, but it seemed that manually creating an adjacency matrix for 1000 regions would be impractical, and a good candidate for an automated approach. I'm just not clear what information such an approach might have to work with. Can the adjacency between two vertices be determined based on their values? If so, one approach might be:
for each pair of labels (A,B): get set of vertices within A get set of vertices within B for each vertex pair (Ai, Bj) if Ai and Bj are adjacent, stop, and mark A and B as adjacent
Cheers, Chris