A CLINICAL COMPARISON BETWEEN CONVENTIONAL AND DIGITAL MAMMOGRAPHY UTILIZING COMPUTED RADIOGRAPHY

Previous work utilizing a specially designed mammographic test object has indicated that the image quality of computed radiography (CR) approaches that of a conventional film-screen technique of equivalent radiographic speed. In an attempt to correlate these results with the clinical situation a subjective rating study of the primary physical parameters of noise, sharpness, and contrast has been conducted. Using a randomized viewing sequence two radiologists experienced in mammography were asked to score these parameters utilizing a 5 point scale. A total of 138 images was used, 62% containing no abnormalities and 38% containing either microcalcifications, masses or both. Additionally, the observers scored their confidence of correct classification, and these results were then used to construct receiver operating characteristic (ROC) curves. The subjective rating of the physical parameters results produced similar results to those obtained using the test object, with CR images providing greater contrast than film. The resolution of CR, however, was lower than for film. Overall the noise levels were similar for both modalities, but the digital images occasionally contained high levels of dust artefacts. In one case these artefacts were mistaken for a cluster of microcalcifications. The test object results and the parameter rating results indicate that CR approaches the performance of film-screen images. Statistically the ROC curves produced indicate CR images to be comparable to the film-screen combination. This may be because the increased contrast compensates for resolution limitations and these findings correlate with our clinical experience.