Published: Fri, January 03, 2020
Health Care | By Cedric Leonard

Google AI model beats radiologists at detecting breast cancer

Google AI model beats radiologists at detecting breast cancer

Fresh speculation on the merits of artificial intelligence have surfaced after Google's health division and several others were found to have detected breast cancer more accurately than human radiologists, according to reports.

The American Cancer Society says radiologists fail to detect cancer in mammograms in around one-fifth of cases and that more than 50% of women who receive screening over 10 years receive a false positive.

"There's enormous opportunity, not just in breast cancer but more widely, to use this type of technology to make screening more equitable and more accurate", Dominic King, the UK lead at Google Health, told the WSJ.

Google AI model was trained and tuned on a representative data set comprised of de-identified mammograms from more than 76,000 women in the United Kingdom and more than 15,000 women in the USA, to see if it could learn to spot signs of breast cancer in the scans.

It was then assessed on a separate data set of over 28,000 British and United States women.

It also unlocks potential for reducing false findings, as compared to human interpretation, the AI showed an absolute reduction in the proportion of cases where cancer was incorrectly identified (5.7%/1.2% in the UK and US data respectively), as well as cases where cancer was missed (9.4%/2.7% in UK/US data). When they disagree, a third is consulted.

The study saw a computer created by Google's AI experts compete with medical experts as they screened mammograms.

FAA proposes rule change to force identification of Colorado, Nebraska drones
However, they do not appear to be malicious, and a drone expert says they appear to be searching or mapping out the area. In the week before Christmas the unmarked drones appeared over Phillips county every night at 7pm.


"What's unusual is that it compares the algorithm to a totally realistic clinical scenario", he adds, whereas past studies have used specially selected mammograms that were analysed in a somewhat artificial setting, for example with a higher-than-usual proportion of cancer cases.

Google's AI has been trained to read and analyze mammograms to detect the presence of cancer cells.

"There will of course a number of challenges to address before AI could be implemented in mammography screening programmes around the world, but the potential for improving healthcare and helping patients is enormous".

The issue, Dr Lehman believes, is current CAD programs were trained to identify things human radiologists can see, whereas AI learns to spot cancers based on the actual results of thousands of mammograms. Lehman states that this might "exceed human capacity to identify subtle cues that the human eye and brain aren't able to perceive".

Although computers have not been "super helpful" so far, "what we've shown at least in tens of thousands of mammograms is the tool can make a very well-informed decision", Etemadi said.

The study has some limitations, however. The findings showed that using the AI in this way could reduce the workload of the second reviewer by as much as 88%, which could ultimately help to triage patients in a shorter timeframe.

Crucially, the team has yet to show the tool improves patient care, said Lisa Watanabe, chief medical officer of CureMetrix, whose AI mammogram programme won U.S. approval a year ago.

Like this: