[Studies on the diagnostic accuracy of lymphography (author's transl)]

Contradictory reports in the literature on the reliability of lymphography stimulated the authors to test the diagnostic accuracy, employing methods which are approximately analogous to practice, using carcinoma of the cervix as the model on which the study was carried out. Using 21 observers it was found that there was no correlation between their experience and on-target accuracy of the diagnosis. Good observers obtained an accuracy of 85% with good proportions between sensitivity in the recognition of detail, specificity and readiness to arrive at a decision on the basis of discriminatory ability. With the help of the concept of the ROC curves, the position taken up by the observers in respect of diagnostic decisions, and a complex manner of assessing the various characteristic factors determining diagnostic accuracy, are demonstrated. This form of test, which permits manipulation of different variants of diagnosis, is recommended, among other things, for performance control at the end of training and continuing education courses in other fields of x-ray diagnosis as well.