Assessment of image quality and total performance in Norwegian mammography laboratories. Findings in a national survey based on different phantoms and ROC methodology.
To assess the image quality at different mammography laboratories.
Two commercial mammographic test phantoms and one phantom based on excised mammary tissue were used in an assessment of the imaging chain and total performance at 45 Norwegian mammography laboratories. The breast-tissue phantom was used for a receiver operating characteristics (ROC) analysis. This was carried out by putting overlays with identifiable regions (some of which contained a cluster of simulated calcifications) on top of the mammary tissue, and then having a radiologist report the confidence of a finding for each region.
The areas under the ROC curves were in general high. In nearly all the laboratories, performance was improved when a magnification technique was applied. There were wide variations among the laboratories in total performance as measured by the area under the ROC curve, and also in the physical parameters derived by means of the commercial phantoms. In general, a good ROC performance was associated with a good physical performance in the imaging chain.