Rick Reynolds pointed out for functional clusters, the interiorosity might be better if it were weighted by the function instead of the mask dataset. Also, it's best to use the smallest sphere that can encompass the whole ROI.

3dLocalstat -nbhd 'SPHERE 15' -stat sum -mask crescentroi+orig. -prefix interiorosity2 func_slim+orig'[1]'

This required a small change to the code to allow for a sum statistic, so this change will be included in the next binary distribution.

I tried to find a similar measure elsewhere, and this paper includes a measure called "interiorness". Although computed differently, I believe it ends up being the same thing.

[www.springerlink.com]