Drug News
Is it time to read screening mammograms with CAD?
By Sue Mueller
Oct 2, 2008 - 10:03:25 PM

If you like the article, could you please do us a favor? Just tell Google News Services that you like foodconsumer.org included in Google News Services. Inclusion in googlenewsservices means many more people can read articles like this. Thanks.
------

Thursday October 2, 2008 (foodconsumer.org) -- With a computer-aided detection (CAD) system, one radiologist can evaluate screening mammograms for breast cancer as effectively as two radiologists without CAD, according to a new study.

The study scheduled to be published in the October 16 issue of the New England Journal of Medicine showed the computer-aided method offered the same sensitivity, specificity, and positive predictive value as the current method.

The current method so called double reading requires two readers or radiologists to read screening mammography commonly in European countries, but not in the U.S., according to Reuters.

Dr. Fiona J. Gilbert, from the University of Aberdeen in the UK, and colleagues demonstrated in their study that single reading with CAD assistance can be as effective as double reading.

The researchers compared the computer aided evaluation with the current double reading for their performance in evaluating screening mammograms from 31,057 women.

The cancer detection rate (sensitivity) was 87.7 percent with double reading compared to 87.2 percent with single reading coupled with computer assistance.   The rate of recalls for further assessment was 3.4 percent with double reading compared to 3.9 percent with computer-aided single reading, a slightly, but significantly higher increase.

The specificity, and positive predictive value for both evaluation protocols are similar, 96.9% and 18.0% for single reading coupled with computer assistance versus 97.4% and 21.2% for double reading respectively.

Some previous studies came to a similar conclusion that CAD use for interpretation of screening mammograms is as beneficial as having a second radiologist review the mammograms.

A prospective study of 4,341 screening mammograms led by Priscilla Slanetz, MD, MPH, at Caritas St. Elizabeth's Medical Center in Boston, MA showed CAD actually increased the cancer detection rate by 5.4 percent.

Slanetz said the increase could be greater in other facilities where mammographers may not be as experienced as those at Caritas St. Elizabeth's.

But the computer aided single reading missed 11 of 38 cancers in Slanetz'a study.   The rate of recalls for further assessment was also found higher with the CAD protocol than the double reading, 13.4 percent versus 11.5 percent.

The study was presented on May 4, 2004 at the American Roentgen Ray Society Annual Meeting in Miami Beach, FL.

But not all doctors are convinced that the CAD plus single reading could beat the double reading.   Group Health Cooperative Center for Health Studies issued a statement on Dec 4, 2006 warning radiologists not to depend too much on the use of computer-assisted detection technology when reading screening mammograms.

The organization cited one of its studies published in the December 2006 issue of the American Journal of Roentgenology and said CAD sometimes missed lesions among other things.

Stephen Taplin, MD, MPH, at Group Health and colleagues analyzed results from two protocols with or with CAD assistance and found CAD assistance increased specificity from 72 percent to 75 percent, meaning CAD increased radiologists' ability to determine that a woman without cancer was cancer-free.  

The sensitivities for both methods were similar; meaning with radiologists using CAD did not increase their ability to determine who have breast cancer.

The problem with the CAD approach is that this technology did not mark all visible abnormalities.

"This means that the radiologists may have been deferring to CAD and believing its interpretation rather than their own interpretation," said Taplin. "This is something the originators of the technology say radiologists should not do. This study shows that it is hard to ignore the technology, and it raises the question of whether there is a potential for CAD to do harm."

"Our study shows that radiologists must continue to rely on their own judgment when determining whether lesions seen on mammograms require further testing," said Taplin.