Interactive, mathematical, and sequential consultative methods in diagnosing renal masses on excretory urograms

Using the clinical setting of diagnosing renal masses on excretory urograms, we compared the diagnostic efficacy and costs resulting from different consultative methods. These included face-to-face interactive consultation, mathematical combination of a fixed number of radiologists' confidence judgments, and computerized sequential decision making. This last method mathematically combines a variable number of individual judgments into an aggregate diagnosis based on diagnostic certainty. Six radiologists interpreted 42 proven urograms individually, with face-to-face consultation for selected cases, and interactively in groups of three. Individual diagnostic probability estimations were mathematically combined for the mathematical and sequential models. All models of consultation resulted in both higher diagnostic accuracy and lower costs than individual radiologists interpreting urograms alone. While the sequential model provided the highest diagnostic efficacy, it was only slightly more accurate than the other models tested. Radiologists interactively consulting on cases they considered difficult was the least costly method, approximating the projected costs of sequential decision making. Interactive consultation among radiologists is a cost-effective practice; sequential decision-making is a promising technique for improving diagnostic efficacy and reducing costs, and further evaluation is warranted.