Observer performance in radiologic interpretation has traditionally been based upon Receiver Operating Characteristic (ROC) analysis. A statistical method that evaluates the reproducibility or consistency of radiologic interpretation is presented. Since this method is not based upon pathologic verification, it cannot be used to assess accuracy of interpretation and is therefore not a substitute for ROC analysis. Rather, it can either supplement ROC analysis or be of use in situations in which the latter is not feasible or practical.
No takes yet. Share an insight, caveat, or question.
Herman et al. (1984) studied this question.