Subdividing BI-RADS category 4 breast lesions observed on magnetic resonance imaging: Is it feasible?