BOSTON (Reuters) - Computer software designed to look for breast tumors can read mammograms nearly as well as an extra pair of human eyes, according to a study published on Wednesday that could change the way the X-rays are read.
In at least 12 European countries, two health specialists read each mammogram, which increases the cancer detection rate by as much as 14 percent. Replacing one of those workers with a computer would cut costs.
Although double-reading is more accurate, it is not widely done in the United States, where single-reading is the standard. Computer-aided detection (CAD) systems are used for only about 25 to 30 percent of U.S. cases, according to the research team, led by Dr. Fiona Gilbert of the University of Aberdeen in Scotland.
“There’s a lot of controversy over the added value of CAD. This study adds weight to the fact that CAD is a viable additional tool,” Gilbert said in a telephone interview.
Gilbert’s team studied the records of more than 31,000 British women. Double-read film mammograms detected 88.7 percent of tumors versus a detection rate of 87.2 percent when a computerized system replaced one of the readers.
The only drawback was that slightly more women were called back for a second look when the computer was used. The recall rate was 3.4 percent for two human readers and 3.9 percent for the combination of person and machine.
The Gilbert team, in a study being published online by the New England Journal of Medicine, also found no difference in the physical attributes of the tumors detected by the two techniques.
The paper did not assess the accuracy rate of new digital mammograms, which eliminate the need to digitize X-ray film so computer software can check for tumors. They also save the reader having to look back and forth between the film and a computer screen.
The study did not weigh the potential cost savings of computer reading.
“The additional costs of the computer-aided detection equipment and the costs associated with an increase in recall must be balanced against the potential savings in reader time. Clearly the cost-effectiveness of computer-aided detection requires investigation,” the researchers wrote.
“Where single reading is standard practice, computer-aided detection has the potential to improve cancer-detection rates to the level achieved by double reading.”
Editing by Maggie Fox and Patrick Markey