I was asked for this about a year ago by a postdoc here. However, I've done exactly nothing (unless you count printing out Schmahmann's paper).
As I understand it, this would require
1) creating a set of markers, somewhat like the Talairach stuff, to "normalize" the cerebellar part of the brain, and then
2) creating a mask volume, like TTatlas+tlrc, to mark the various regions
#1 is possibly doable by me, but #2 is harder (for a non-brainologist, at least).
bob cox