Why Most Women Reject AI to Interpret their Mammograms

Why Most Women Reject AI to Interpret their Mammograms

by October 16, 2020

Artificial Intelligence

Reason behind Women rejecting AI for their Mammography

Artificial intelligence (AI) algorithms are designed to interpret mammograms independently, which might be obtaining ground when it comes to radiologist conference, although the public-at-large does not share this sentiment. Roughly 80% of women say they oppose letting artificial intelligence interpret their studies without a radiologist being directly associated somehow.

A study published on October 12 by the Journal of the American College of Radiology, a team of investigators from the University of Groningen, Netherlands reveals that a majority of women (78%) are not comfortable with AI alone analyse their mammograms for signs of breast cancer.

Yfke Ongena, PhD, Assistant Professor with the University’s Centre of Language and Cognition, says, “Despite recent breakthroughs in the diagnostic performance of AI algorithms for the interpretation of screening mammograms, the common people currently do not support entirely independent use of such systems without involving a radiologist.”

The response that shows AI can outperform radiologists is surprising. It also shows that radiology as an industry should force the breaks on implementing any artificial intelligence algorithms as the sole interpreter of mammograms. It could also work to educate women about the capabilities and efficacy of these algorithms. Ultimately improving people’s perception about AI is vital because the industry currently is lacking enough radiologists in many regions, even though they are qualified to interpret mammograms, the team points out.

The survey result also reveals that when it comes to using AI as a second read on select studies, 42% disagreed, 31% agreed, and 27% were undecided.

Ongena’s team conducted a survey of 922 women aged between 16 and 75 in two waves. The picked women aged between 16 and 40 because they represent the next generation of women who will undergo routine screening mammography. They collected responses from December 2018 first, and then from April 2020. The team asked women how they feel about the necessity of a human check AI as a selector for second reading, and if developers or radiologists were responsible for errors.

They aimed to focus on how women feel about having artificial intelligence interpret their mammograms in light of how the conventional press has portrayed these tools in a broader positive manner. Initially, they postulated that responses from women would be in alignment.

They proved to be wrong based on the survey result. Although the team concluded, there is room for AI to acquire acceptance as a second reader that supports the radiologist. Only 17% of women practically opposed using the tool in association with her radiologist.

Ongena claims, “The combination of a radiologist as a first read and an AI system as a second reader seems to be the most acceptable approach to the population at present, though still not fully adopted by the entire population.” They added, “Improved information supply and education about the development, possibilities, and limitations of AI algorithms in screen mammography may potentially get through some of the perceived hurdles and increase acceptance of this new technique in clinical practice.”

The most crucial step to enhance the acceptance of AI in the general population, they say, is bolstering patient education about the possibilities of the technology. Overall, women who opposed the use of AI also reported having a lower level of education.

The study does have a significant limitation as survey responses were received from a country where mammography is routinely offered to women age between 50 and 75, and at least two radiologists interpret those scans. Other countries might lack the required resources to process mammograms at this level. In the case of the U.S, only Medicare screens breast cancer routinely.

It is critically important to consider women’s opinions of these AI,” the team opines.

Ongena says, “The voice of the population who will undergo AI-based diagnostic tests is pivotal in this context because it is a determining factor for the boundaries within which an AI system is allowed to operate.”

They add that the willingness of subjects to participate matters for a successful breast cancer screening program, and it might be impacted if artificial intelligence algorithms are used without taking into account the people’s wishes, concerns, and objections.