Decision aids for improved accuracy and standardization of mammographic diagnosis

This paper examines the gains in the accuracy of mammographic diagnosis of breast cancer achievable from a pair of decision aids. Twenty-three potentially relevant perceptual features of mammograms were identified through interviews, psychometric tests, and consensus meetings with mammography specialists. Statistical analyses determined the 12 independent features that were most information diagnostically and assigned a weight to each according to its importance. Two decision aids were developed: a checklist that solicits a scale value from the radiologist for each feature and a computer program that merges those values optimally in an advisory estimate of the probability of malignancy. Six radiologists read a set of 150 cases, first in their usual way and later with the aids