Simplified MRI prediction of clinically definite multiple sclerosis: a stepping stone towards treatment criteria?