Google Health AI beats radiologists in breast cancer detection

January 2nd
A radiologist reviewing mammograms. (Rex/Shuttershock)

Google Health has developed an image recognition AI to help radiologists detect warning signs of breast cancer.

The AI, which outperformed experts in the US and UK, may go on to "have a role in aiding the early detection of breast cancer," according to Nature.

The team behind the program included researchers from Google Health and Imperial College London, who trained the AI on past mammograms from women in the UK and US. After training, the AI was tasked with assessing 30,000 other mammograms, some of which belonged to women who were diagnosed with cancer, and others to those who were cancer-free for a year after the screening.

In testing, the AI program outperformed a US-based professional panel by 5.7% in false positives and 9.4% in false negatives. In the UK, it produced 1.2% and 2.7% fewer false positives and false negatives, respectively.

The program may help understaffed hospitals in the UK, according to The Guardian:

"Like the rest of the health service, breast imaging, and UK radiology more widely, is understaffed and desperate for help," said Dr Caroline Rubin, vice-president for clinical radiology at the Royal College of Radiologists. "AI programs will not solve the human staffing crisis, as radiologists and imaging teams do far more than just look at scans, but they will undoubtedly help by acting as a second pair of eyes and a safety net."

The AI requires more development, researchers say, and will have to pass clinical trials before it can be released.