prove on conventional imaging methods by allowing the different tasks of image acquisition, processing, and display to be separated and therefore refined, as well as allowing adjunct technologies such as CAD to be used more easily.59 But despite the introduction of full-field digital mammography units into the market, the question of whether digital actually improves cancer detection rates or workflow remains open.

Digital mammography systems offer better contrast and lower spatial resolution at a lower radiation dose than traditional screen film mammography.59 The relative diagnostic accuracy of digital mammography compared to traditional mammography is still undergoing study through the large Digital Mammographic Imaging Screening Trial (DMIST). As of this writing, digital mammography appears to improve specificity—possibly due to the flexibility of image display available to the interpreting radiologists.31 Digital units probably also improve workflow, allowing radiologists to view images in less than a minute, compared to the 8 to 10 minutes required from screen film systems.

Because digital mammography devices are more expensive than conventional devices, they will have to offer substantial advantages over film-screen mammography in order to be widely used. Research to date has not shown a dramatic difference between the two techniques. In a 2001 study, 4,945 women had both conventional and digital screening mammography exams. The conventional mammography device found a few more cancers than the digital unit; both devices missed cancers the other found. Overall, there were no major differences in cancer detection rates between the two techniques, although if the digital technique had been used alone, recall rates would have been lower.70 Similar results were reported in 2003 from a Norwegian study of 3,683 women.106 Each woman in the study had both digital and screen film mammography exams and they were independently interpreted. The cancer detection rates for the two imaging modalities were not significantly different, although the recall rate was slightly higher for digital mammography and the positive predictive value based on needle biopsy was slightly higher for screen film mammography.

In 2001, the National Cancer Institute (NCI) and the American College of Radiology Imaging Network (ACRIN) launched the multicenter DMIST study to compare digital mammography with standard mammography for the detection of breast cancer. The 49,520 women enrolled in the study will be followed for one year after receiving both digital and conventional mammograms. For further discussion of ACRIN and DMIST, see Chapter 6.

Computer-Aided Detection

Since 1989, technology developers have conducted experiments on the use of computer capability to aid in reading radiological images. Because



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement