CAD: The Latest Chapter in the Breast Cancer Detection Debate

I’ve been trying to sort out the latest findings about computer-aided detection, or CAD, for mammography.

A new study, reported in the New England Journal of Medicine last week and then in the popular media (Click to read stories published on MyBreastCancerNetwork.com, in the New York Times, Washington Post), found that CAD, a highly touted, highly expensive computerized system for examining mammograms is leading to less accuracy, not more, in breast cancer detection. That was not the anticipated outcome of the new technology.

First of All, What is CAD? And, How Does It Work?

A computer-aided detection system is a computer that is combined with software to find suspicious spots on a mammogram and mark them. A radiologist reads a mammogram. Then the radiologist double-checks his or her conclusion with the computer system to see if anything was missed. Here’s where things can get a little dicey. CAD mammography often finds harmless spots that lead to false scares. And a good, experienced radiologist may do as just as well or better at spotting cancers as the computer systems, the new study suggests.

So is the CAD equipment worth the money?

Computer-aided detection systems are expensive, ranging from $50,00 to $175,000. Several companies sell the detection systems, which are used in about 30 percent of mammography centers. Medicare pays an extra $20 for every mammogram read with CAD, assuming that the equipment would find cancers that radiologists missed and would save lives. The new study of CAD analyzed 429,345 mammograms, taken from 1998 to 2002, at 43 mammography centers. In that time period, seven of the centers switched to CAD.

Thus, the investigators compared results, with and without the software, to help radiologists find suspicious spots. The researchers’ conclusion: CAD led to “significantly higher” rates of false positives and biopsy rates and “significantly lower” accuracy. Many women—31 percent—were called in for additional tests and 20 percent more had biopsies. Yet the only thing CAD seemed to detect more of is a precancerous condition called ductal carcinoma in situ (DCIS), which is often harmless, although it can lead to invasive breast cancer. According to some doctors quoted in media reports of the study, a test that finds more DCIS and less invasive cancer may be doing more harm than good.
 
So does the new study mean that CAD is no good?

If only things were that simple. According to a letter posted on Aunt Minnie, a popular web community for radiologists, technologists and hospital administrators seeking to purchase and understand imaging technology, more research is needed before everyone writes off the CAD mammography as a big waste of money. The letter, written by Brian Casey, the editor-in-chief of the Web site, argues that it’s too soon to give up on the CAD technology. “Instead, they (CAD proponents) should redouble their efforts to improve the technology and its application in breast imaging,” he says.