Hi,

I suspect there is a better/easier/obvious way to do the following, any feedback will be appreciated.

I am using 3dfractionize to transform a +tlrc dataset with n distinct nonzero values (one voxel each, or input has n nonzero voxels) into the +orig space of 33 distinct subjects, using the -vote option to preserve the values, and the -warp option to provide the transformation. If the -clip option is not invoked, then more than n nonzero mask voxels are written to the dataset in each subject's +orig space. The clip level such that exactly n voxels are preserved varies between subjects; there doesn't seem to be a "common" clip level for all subjects which will preserve all n voxels but not more.

I would like to find the clip value for each subject such that the number of distinct nonzero values in the input is equivalent to the number of nonzero mask vozels written to the dataset ( sclip | nvote_val == ijk ) . How can I automate this search? Or, perhaps I should be trying something else entirely.

Thanks a lot,
Clara.