Google is developing an AI tool to better detect breast cancer and it has been successful so far, based on a research paper that was published yesterday.
Breast cancer can be detected through a mammogram — an X-ray diagram of the breast — which is further examined by the doctor. Based on the mammogram, doctors can discover symptoms of mammograms, sometimes much earlier before women can feel it when they examine themselves. However, it is not easy to analyze these X-ray images which can lead to many false negatives and false positives.
Google’s Health division has taken this issue into their hands by working with several clinical partners — Deepmind, Cancer Research UK Imperial Centre, Northwestern University and Royal Surrey County Hospital — and publishing their initial findings that reveal the AI tool to have greater accuracy compared to human experts.
Using data sets of more than 28,000 women from the U.K. and the U.S., the AI model managed to outperform the human experts despite having less information about the patient (histories and previous mammograms). The model reduced false negatives by 9.4% in the U.S. and 2.7% in the U.K., while the false positives were reduced by 5.7% in the U.S. and 1.2% in the U.K.
However, this doesn’t mean that Google is trying to create something that can replace the radiologists. Google researcher, Shravya Shetty told The Verge, “There are a number of cases where the radiologists catch something that the model misses, and vice versa. Bringing the two together could strengthen the overall results.” Google is simply trying to create a tool that can help radiologists make a better judgment, reducing misdiagnosis and increasing efficiency.
