Liven scintigraphy for detection of metastatic dis

ease has been a widespread nuclear medicine procedure for 20 years. During this time, numerous reports of the diagnostic worth of this test have been published (1—38), with reported sensitivity values ranging from 60 to 98%, and specificity values from 64 to 99%. Although many series have consisted of a small number of patients, and consequently may have large uncertainties, several ne ports dealing with hundreds of patients each have been in substantial disagreement. In medical imaging, the concept of sensitivity and specificity as measures of performance was introduced in 1947 by Yerushalmy (39). These measures were de fined respectively as the probability of detecting a spe cific disease entity in a patient who has the disease, and the probability of ruling out the disease in a patient who does not have it. The important insight in this paradigm of performance is that there are two separate classes of patientsin whichthe probabilityofcorrectdiagnosisand the consequences of error are markedly different.