Towards taking the guesswork (and the errors) out of diffusion tractography