I am trying to do a within-subject co-registration between b0 volumes and T1's for my subjects. This is an old dataset that does not have anything for correcting b0 inhomogeneities. The registration using --dti and normal parameters for bbregister led to "okay" within-subject registration. I am wondering if modifying the --dof (like 12) option for bbregister or applying some type of linear/nonlinear registration might improve the registration?