Assessment of breast lesions on stereoscopic and monoscopic digital specimen mammograms: an ROC study

An ROC study was conducted to evaluate the usefulness of assessing breast lesion characteristics with stereomammography. Stereoscopic image pairs of 158 breast biopsy tissue specimens were acquired with a GE digital mammography system. Two stereo image pairs were taken at 1.8X magnification geometry and at approximately orthogonal orientations for each specimen. Display software was developed for a high resolution MegaScan CRT monitor driven by a DOME stereo display board. The specimens contained either a mass, microcalcifications, both, or normal tissue. About 40% of the specimens were found to contain malignancy by pathological analysis. Five MQSA radiologists participated in the observer performance experiment. The two views of each specimen were read independently and were separated by a large number of other specimen images to reduce any effects of memorization. Each observer read 316 specimen images in a systematically randomized order. The observer first read the monoscopic image and entered his/her assessment in terms of the confidence ratings on the presence of microcalcifications and/or masses, margin clearance, BI-RADS assessment, and the likelihood of malignancy. The corresponding stereoscopic images were then displayed on the same monitor and were viewed through stereoscopic LCD glasses. The observer was free to change the ratings in every category after stereoscopic reading. The ratings of the observers were analyzed by ROC methodology. For the 5 MQSA radiologists, the average Az value for estimation of the likelihood of malignancy of the lesions improved from 0.70 for monoscopic viewing to 0.72 (p