AI Breast Cancer Screening: Not there yet

When used to detect breast cancer in mammography screenings, AI algorithms unfortunately did not do a good job relative to medical professionals.

In nearly all cases, they are less accurate than a radiologist based on a systematic review conducted by UK researchers and published in the peer-reviewed medical journal, The BMJ.

Why is this notable?

Tech media and AI advocates have looked at radiology as one of the medical fields most susceptible to AI automation.

Looking deeper into AI breast cancer screening studies

University of Warwick researchers looked at 12 studies conducted between 2010 and 2021 looking at the accuracy of AI in detecting cancer in digital mammograms. In total, they screened 130,000 women.

The results

Out of the 36 AI systems, 94% were less accurate than a single radiologist.

And 100% of the AI systems were less accurate than two or more radiologists working together.

When the researchers looked at five smaller studies — totaling roughly 1,000 women’s screenings — these found that AI systems bested radiologists. But they found that these results didn’t hold up in more comprehensive reports.

The researchers wrote:

“Current evidence for AI does not yet allow judgment of its accuracy in breast cancer screening programs, and it is unclear where on the clinical pathway AI might be of most benefit.”

Bottom line

AI is not the golden solution to everything.

This is especially true when it concerns things that are inherently complex and difficult to automate and things that literally involve life-or-death (e.g., autonomous vehicles, health and medical matters).

When used in medical applications, algorithmic AI systems need to have strict oversight — for blind spots, bias, and potential harms it can cause.

Related Posts