[Freesurfer] Question about hippocampal-subfield probability map

Dear experts,

The hippocampal-subfield are expressed as probability maps. For each subfield, each voxel is assigned a probability of inclusion on a scale of [0-255]=[0-1]. However, If I were to select only voxels from each subfield's probability map corresponding to a >50% chance (i.e. mri_binarize -min 150), then I would expect that no subfield would overlap with another (i.e. a voxel shouldn't have greater than 50% probability of belonging CA1, AND a greater than 50% probability of belonging to CA4-DG. However it appears that this is not the case, and overlap does occur. Thus, unless I messed up, subfields are not mutually exclusive. Have I got this right? And if so, how do I obtain binary maps of subfields with no overlap?

Thanks

Josh