Comparison of Objective and Subjective Methods to Assess Imaging Performance in Digital Mammography

Digital images were acquired of an anthropomorphic breast phantom (Rachel) at mAs values that ranged from 40 to 120 mAs. Digital mammograms were obtained with/without added lesions, which permitted the generation of a digital version of the lesion alone. The lesion was added back to the digital image of the Rachel phantom at varying levels of contrast. A subjective method of lesion visibility was compared to an objective method used Four Alternate Forced Choice (4AFC) methodology. Both methods showned observer performance for detecting a mass lesion to be independent of x-ray tube output between 40 and 120 mAs. The objective approach results in a higher precision, but alo requires many more images.