Mammography AI can cost patients extra. Is the Extra Cost Justifiable?

As I registered for my annual mammogram at a Manhattan radiology clinic in November, a surprising offer was presented by the front desk staff. For an additional $40, I could opt for an artificial intelligence analysis of my mammogram, a service not covered by insurance. Uncertain about the value, I declined, prompting reflections on whether this enhancement was necessary for routine screening. Does it signify a deficiency in the accuracy of traditional mammograms? And why doesn’t insurance cover this apparently promising AI analysis?

This scenario isn’t unique to me. A colleague’s mother faced a similar proposition at a suburban Baltimore clinic, complete with a persuasive pink pamphlet extolling the virtues of AI in mammography for the same price tag of $40. She, too, declined the offer, echoing the collective skepticism.

In recent years, AI software designed to assist radiologists in detecting issues or diagnosing cancer through mammography has entered clinical use. This software can process and analyze vast image datasets, identifying patterns and abnormalities that may elude human radiologists. While it holds the potential to significantly improve the detection of suspicious breast masses and facilitate earlier breast cancer diagnoses, questions linger about its routine use in regular clinical practice.

Despite promising studies showcasing improved detection rates, some radiologists emphasize the need for further research and evaluation before embracing these tools as standard practice. Etta Pisano, Chief Research Officer at the American College of Radiology, acknowledges the promise but calls for more information to determine individual benefits.

The radiology clinics under discussion are part of RadNet, a company operating over 350 imaging centers across the country. Gregory Sorensen, the company’s Chief Science Officer, points to research involving 18 radiologists, revealing improved performance using AI across the board. However, the question remains: Is the technological analysis worth the additional cost for patients?

SEE ALSO: The Impact of AI in Healthcare Diagnostics: Balancing Accuracy and Potential Biases

Laura Heacock, a breast imaging specialist at NYU Langone Health’s Perlmutter Cancer Center, acknowledges that some individuals may find reassurance in AI-enhanced mammograms due to heightened anxiety. Still, she emphasizes that expert human interpretation remains the standard of care.

With roughly two dozen AI products authorized by the FDA for cancer detection in mammograms, the absence of billing codes for AI interpretation poses a challenge. While RadNet reports that 35% of women opting for mammograms at their facilities pay for the additional AI review, practices vary in handling payment for AI mammography.

Constance Lehman, a professor of radiology at Harvard Medical School, stresses that charging patients for AI analysis may undermine equity, limiting access to those who can afford it. RadNet’s goal is to cease patient charges once health plans recognize the screening’s value.

Large trials are underway in the U.S., with some research indicating higher cancer detection rates with AI. However, questions persist about the generalizability of these findings and the need for more diverse training sets across different demographics.

The echoes of past disappointments with computer-assisted mammography from the late 1980s and early 1990s linger, but proponents argue that today’s AI tools are more sophisticated and hold substantial promise. Nonetheless, concerns about affordability and equitable access to these advancements remain, prompting a critical evaluation of the role of AI in shaping the future of mammography.