Computer-Aided Mammography Finds More Cancer, More False Positives

Ruzanna Harutyunyan's picture

Computer programs designed to help radiologists could identify more cases of breast cancer, but they might also increase the number of false positive results, which can lead to biopsies in healthy women, according to a recent systematic review.

Using computer-aided detection (CAD) mammography, “you do catch some cases that would have been missed if the mammogram had been read only by a single radiologist,” said review author Meredith Noble, a research analyst at ECRI Institute.

Typically, a radiologist examines a woman’s screening mammogram to check for signs of cancer. When using CAD with mammography, the radiologist still reads the mammogram, but a computer program also evaluates the mammogram and marks suspicious areas for the radiologist to review further.

Investigators led by Noble synthesized data from seven previously published studies of CAD mammography’s use in 392,015 healthy women with no lumps or other symptoms of breast cancer.

The current report, released this month, is an update of an earlier review also published by ECRI Institute, a nonprofit health services research agency that produces systematic evidence reviews on medical devices, drugs, biotechnologies and procedures.

Investigators found that in women with no symptoms, screening with CAD mammography identified an estimated 84.2 percent to 87.6 percent of women with cancer, a finding referred to as test sensitivity.

In addition, an estimated 88.1 percent to 88.3 percent of healthy women correctly received negative test results when undergoing CAD mammography, a finding referred to as test specificity.

Researchers compared CAD mammography results to biopsy results and patient follow-up records to determine whether if the women received a breast cancer diagnosis within a year after the mammogram.


The updated evidence review indicated that CAD mammography helped identify an estimated 50 additional cancer diagnoses for every 100,000 women who underwent the screening test, compared to having the mammogram read only by a radiologist.

The analysis also estimated that following CAD mammography, between 1,090 and 1,290 women per 100,000 healthy women — women who did not have breast cancer — would be recalled for further testing in the form of more imaging studies or biopsy based on abnormal mammogram results.

ECRI Institute estimated that 80 of those women who had false-positive results would undergo biopsy.

Joshua Fenton, M.D, a researcher who focuses on breast cancer screening, said a concern “is that we may have variability in how doctors are using this technology.” He said that some providers might simply rely on CAD to catch potentially cancerous areas, instead of interpreting the mammogram first and using CAD as secondary analysis tool.

“That might result in a different outcome in the community,” Fenton said.

Fenton, an assistant professor at the University of California-Davis, has no affiliation with the ECRI review.

Based on the evidence review, Fenton agreed with the authors that there are not enough data to say for sure that CAD has clear public health benefits, that it improves breast cancer mortality or that it helps doctors detect the most dangerous breast cancer.

Although the results of this evidence report pertain mainly to health care providers, patients do need to know about the advantages and potential drawbacks of CAD mammography, Noble said. Most notably, physicians should tell women who undergo CAD mammography screenings that there are more false-positives when CAD is used.

False-positive findings can provoke anxiety, but some women and their doctors might find them to be an acceptable trade-off for identifying some additional cancers, Noble added.

CAD mammography received approval by the U.S. Food and Drug Administration in 1998.