Artificial Intelligence in Mammogram Analysis: A New Tool in Breast Cancer Detection

Artificial Intelligence Aims to Improve Breast Cancer Detection in Mammograms

Clinics around the country are beginning to offer patients a new service: having their mammograms read not only by a radiologist but also by an artificial intelligence model. The hospitals and companies providing these tools claim they can expedite the work of radiologists and detect cancer earlier than standard mammograms alone.

Mammograms currently identify approximately 87 percent of breast cancers but are more likely to miss cancer in younger women and those with dense breasts. They can also lead to false positives, requiring additional testing, and may detect precancerous conditions that may not pose serious risks but still lead to unnecessary treatment.

Dr. John Lewin, chief of breast imaging at Smilow Cancer Hospital and Yale Cancer Center, acknowledges that mammography is not a perfect science.

Experts are optimistic about the potential to enhance the accuracy of breast cancer screening, as 300,000 women are diagnosed with breast cancer each year in the United States. However, they have concerns about whether A.I. tools will effectively work across diverse patient populations and significantly improve breast cancer survival rates.

How does A.I. analysis work?

Mammograms contain a wealth of information on breast tissues and ducts, with certain patterns indicating potential signs of cancer. A.I. models can identify patterns that may be challenging for humans to differentiate from normal breast tissue, helping radiologists prioritize scans and detect suspicious areas that require further attention.

In a large study in Sweden, an A.I. model improved breast cancer detection by 20 percent, picking up more cases of cancer compared to radiologists. Additionally, a Danish study in 2022 showed that an A.I. model reduced false positives, minimizing the need for additional testing after a suspicious spot is identified in a mammogram.

Despite these promising results, it remains unclear whether A.I. analysis will reduce breast cancer-related deaths or simply increase early cancer detection rates. Radiologists are uncertain about how well European findings will apply to the U.S. population and whether A.I. models can effectively work across diverse racial and ethnic groups.

Is it worth paying for an A.I. mammogram?

The Food and Drug Administration has authorized around two dozen mammography A.I. products, but there is currently no billing code for radiologists to charge insurance providers for this technology. Some clinics may pass the cost to patients, while others may offer A.I. analysis for free or keep it for research purposes until its value is certain.

While A.I. may not be necessary for all mammograms at this time, it could provide additional reassurance for patients who are particularly anxious about their results. As A.I. technology continues to evolve and become more integrated into routine care, insurance companies may consider reimbursing the cost of A.I. mammograms in the future.