Hi, if I understand correctly, you are trying to get a 10mm binary spherical mask (ROI) centered around a given point. I just tried your approach, and I think the trouble you're having is floating point error, given that you end up working with
tiny values that are quite close to zero. When you open it in fslview, it defaults to an extremely narrow intensity range.
For your purpose, I would suggest to use maximum dilation rather than mean filtering (although it IS a bit slower), and to use integer output. E.g.:
fslmaths /Users/fg00/fsl/data/standard/MNI152_T1_1mm -mul 0 -add 1 -roi 81 1 141 1 127 1 0 1 -kernel sphere 10 -dilF rSFG10mm -odt int
That will leave the output as 0's and 1's, rather than near-zeros (e-13) and 0.00024 in the mask.