Investigation of diagnostic and image quality attributes: comparison of screen-film to CR mammography

Abstract

Digital mammography is advancing into an arena where analog has long been the gold standard. Direct digital systems may not be the favored solution for a particular site while computed radiography (CR) mammography, remains unproven worldwide. This pilot study responds to the growing desire to acquire and display digital mammographic images by exploring the acceptability of CR mammography. Images representing a range of breast tissue types were collected from 49 subjects (17 screening; 32 diagnostic) at four clinical sites. Comparison views were collected on the same breast, under the same compression, using automatic exposure control on state-of-the-art film systems followed by CR. CR images were processed and printed to a mammography printer for hard copy feature comparison. Each image pair in the study was evaluated according to 13 image quality attributes covering noise, contrast, sharpness, and image quality in the overall captured images as well as in each of several particular breast regions (periphery and skin-line, parenchyma and fatty tissue). A rating scale from 1 to 5 was used (strong preference for film=1, strong preference for CR=5). Twelve experienced mammographers at four clinical sites scored a subset of the 49 cases for a total of 64 image pair readings. There were 64 ratings for each of 13 image quality attributes for all cases and, an additional series of scores (four or five attribute ratings) for each abnormality in the category of mass, architectural distortion and microcalcification, for a total of 1069 scores. Based on the pilot study results, it was suggested that CR was equivalent or preferred to conventional screen-film for overall image quality (38% scored 3; 46% scored >3), image contrast (27% scored 3; 59% scored >3) and sharpness (28% scored 3; 50% scored >3). No preference was found when assessing noise. This pilot study also suggested that diagnostic quality was maintained in assessing abnormalities for attributes necessary to evaluate masses and microcalcifications as compared to screen-film.