With the old dcm2nii it would convert a 3D acquisition (e.g. 't1.nii') and then losslessly apply orthogonal rotations so the image slices are roughly axial ('o') and then apply a crop ('co') to remove excess neck. The 'co' opften worked better with the contemporary SPM8 and FSL's BET. 

You can get a similar cropped image if you use dcm2niix's '-x y' option. However, recent FSL BET with the "robust" (-r) option and SPM12's clever built-in spm_maff8 function tend to be much more robust to excess neck signal. I doubt this is required for modern tools