For the third consecutive year, studies comparing digital and film mammography have shown them to be largely equivalent in their ability to detect breast cancer. At the 2001 RSNA meeting, however, researchers examining some of the fine points of digital imaging found that the newer technology may offer other advantages.

Two sessions at the meeting compared breast imaging done with digital and film systems. One study compared the two in trial and clinical settings and found them equivalent in most regards, including sensitivity and specificity. But other presentations found that digital mammography produces more consistent results, may have an advantage in spotting microcalcifications, may result in lower dose exposures, and may be able to spot smaller lesions.

Two observers of the session commented that the results illustrate the difficulties digital mammography faces in overcoming film imaging: limited data and inherent variability among mammography readers.

"We have a paradox. All the technical testing shows digital to be at least as good as film-screen, especially for thicker breasts, or better. Yet in clinical situations, digital is equivalent to film-screen," said R. Edward Hendrick, Ph.D., a research professor at Northwestern University. "It will require very large studies to possibly unravel the small difference between digital and film-screen."

Dr. Robert Schmidt, an associate professor of radiology at New York University and co-moderator of the session, concurred with Hendrick's assessment.

"All the physics is still mapping out the theory that digital will improve things, but when we get to clinical practice we're not seeing it," Schmidt said. "The observer is the biggest variability. Trying to measure these differences, which are not enormous, is still quite difficult."

Despite the research challenges, some studies found advantages for digital mammography. A phantom study conducted at Northwestern University compared the performance of 18 of the first full-field digital mammography (FFDM) systems introduced into clinical practice with the performance of screen-film mammography (SFM) systems already in clinical use.

When used in standard mode, digital exposure times were longer for thin to intermediate breasts, but significantly shorter for thicker breasts. Contrast detail scores were significantly higher for digital than for film. Finally, variability for the digital systems was far lower than for the film systems.

"Results indicate that clinical use of FFDM should improve lesion detection while reducing motion and breast dose for thicker breasts. FFDM provides less variability than SFM in exposure time, breast dose, and image quality," the researchers concluded.

A study by French radiologists determined that digital mammography can detect smaller lesions than film. Using a phantom, they found that contrast detail improved 37% for 5-cm breast thickness and 20% for 8-cm breast thickness.

"This result means that FFDM detected on average 37% and 20% smaller lesions with the same contrast, or 37% and 20% lower contrast lesions of the same size in comparison to SFM for 5-cm and 8-cm breast thickness, respectively," they said.

German researchers compared a screen-film technique for detecting microcalcifications that magnifies images seven times with digital imaging and screen-film imaging magnified 1.8 times. Nearly 10,000 microcalcifications were counted. The 7x magnification technique revealed 4685 microcalcifications. Digital was second at 2654, and film was third at 2366. The differences were statistically significant and indicated an advantage for digital over screen-film imaging, said Dr. Susanne Diekmann of the University Hospital Charite, Berlin.

One of the issues for digital mammography may be a steep learning curve, according to Schmidt.

In a retrospective study, researchers in Oslo, Norway, compared film and digital systems with soft-copy reading in nearly 3700 exams of women who participated in a screening program. Thirty-one cancers were found, 27 by film and 23 by digital. The difference between cancers detected only with film (seven) and only with digital (three) was not significant. The digital system had a slightly higher recall rate (4.6% versus 3.5% for film) and a slightly lower positive predictive value (40.4% versus 45.6% for film).

"A slight difference in the cancer detection rate and the retrospective analysis of malignant features might be due to a learning curve effect," the researchers said.