Application of case series review results to the evaluation of individual cases in diagnostic radiology.

Probability theory provides a simple method for physicians to use their "intellectual linkages" to their past clinical experience in making current diagnoses. Only a pencil and paper are required for making a few likelihood calculations. To illustrate this method, evaluation was done of a new diagnostic sign (presence of knee ossification centers) for differentiating rubella from cytomegalovirus infection in young infants. Two practical questions can be answered by use of this method for calculating probabilities: (1) How certain can one be about either diagnosis when centers are present or absent? (2) How can other radiologists apply these results to their individual cases?