Dear Henry,

The MMRR-21 optimal average template (www.mindboggle.info/data) was constructed using ANTS from the MMRR dataset on NITRC (http://www.nitrc.org/search/?type_of_search=group&q=bennett&sa.x=0&sa.y=0&sa=Search).  No tissue priors exist from these brains for this template, and I don't think anyone has manually segmented the tissue classes in the individual brains.  However, there are automated algorithms such as FAST in FSL and Atropos in ANTS that create a tissue class probability map for a given individual -- maybe you could try running this on the template itself?  You could also register the automatically segmented tissue classes in each of the individuals to the template and create tissue class priors from the coregistered group.

Cheers,
@rno