Google soon to beat humans for spotting the breast cancer with AI technology

Google is currently working on the development of artificial intelligence technology for helping the doctors in early detection of Breast Cancer as per a research paper published in Nature. The AI model will work by scanning the X-ray images that are called Mammograms to reduce the false negatives by almost 9.4 percent. This better negation is going to be a great step in better tests of breast cancers which at the time are missed by 20 percent.

In the world of today, the 2^nd leading cause of deaths among the female population is Brest Cancer and it eats the lung cancer in its mortality rate as well as its overall prevalence. One of the best courses of action to treat Breast cancer is to have the disease detected in the early stages. However, despite the seriousness and dread of the Breast Cancers, they are still only detected with a detection tool called Mammogram but they tend to miss a lot of breast cancer cases.

Shravya Shetty, the researcher working at Google who had co-authored the Nature’s paper, said, “Mammograms are very effective but there’s still a significant problem with false negatives and false positives.”

A study is funded by Google in which researchers are making use of the anonymized mammograms of 25000+ females from the UK and 3000+ females from the US. Shetty says, “We tried the same principles that a radiologist might follow.”

The research team first trained the AI for scanning these X-ray images then look for the breast cancer signs by looking at the changes in the breasts of the 28000+ females. Then the researchers compared the AI’s guesses with the actual medical tests of the females.

The results was astonishingly great with false-negative having a reduction of 9.4 percent and the reduction of false positive by almost 5.7 percent. Two radiologists double-checked the results and the false negative for UK females was cut down by almost 2.7 percent and false positives were reduced by 1.2 percent.

A scientist at Google who was a co-author of the Nature article, said, “This AI model works better than the individual radiologist in both the US and the UK.”

This AI system was still not perfect as there were instances where the model missed the cancer flags that it had missed. A researcher and the co-author of Nature’s paper, Mozziyar Etemadi said, “Sometimes, all six U.S. readers caught cancer that slipped past the AI and vice versa”.

This AI breast cancer detection project is part of Google’s ongoing plans to expand into the field of healthcare. This year, they also partnered with a company named Ascension to get access to the health records of millions of American citizens.