{"id":198903,"date":"2020-01-02T21:49:31","date_gmt":"2020-01-02T13:49:31","guid":{"rendered":"https:\/\/www.geek.com\/featured\/-1815131\/"},"modified":"2020-01-02T21:49:31","modified_gmt":"2020-01-02T13:49:31","slug":"google-health-ai-spots-breast-cancer-human-doctors-miss","status":"publish","type":"post","link":"https:\/\/itteacheritfreelance.hk\/test\/wordpress\/2020\/01\/02\/google-health-ai-spots-breast-cancer-human-doctors-miss\/","title":{"rendered":"Google Health AI Spots Breast Cancer Human Doctors Miss"},"content":{"rendered":"
\n\t\t\t\t\t<\/p>\n
Cancer Cells (Photo Credit: Pixabay)<\/div>\n<\/figure>\n

Artificial intelligence can improve the detection of breast cancer.<\/span><\/p>\n

At least, Google’s model can.<\/span><\/p>\n

Described in a paper<\/a> published by the journal <\/span>Nature<\/span><\/i>, the AI spotted breast cancer in screening mammograms with greater accuracy, fewer false positives, and fewer false negatives than human experts.<\/span><\/p>\n

“This sets the stage for future applications where the model could potentially support radiologists performing breast cancer screenings,” Google Health technical lead Shravya Shetty and product manager Daniel Tse wrote in a blog announcement<\/a>.<\/span><\/p>\n

Created in collaboration with DeepMind, Cancer Research UK Imperial Center, Northwestern University, and Royal Surrey County Hospital, the machine was trained on de-identified mammograms from more than 91,000 women in the US and UK.<\/span><\/p>\n

When put to the test on a separate data set of 28,000-plus ladies, it produced a 6.9 percent reduction of false positives and 12.1 percent reduction in false negatives.<\/span><\/p>\n

“We also wanted to see if the model could generalize to other healthcare systems,” Tse and Shetty explained.<\/span><\/p>\n

So, the team trained their system only on the data from women in the UK, then evaluated it on the data from women in the US. The experiment yielded a 3.5 percent reduction in false positives, and an 8.1 percent reduction in false negatives.<\/span><\/p>\n

This shows the program’s “potential to generalize to new clinical settings while still performing at a higher level than experts,” according to Google.<\/span><\/p>\n

That’s even more impressive when you realize the algorithm was working with less information than human doctors.<\/span><\/p>\n

Medical specialists\u2014in line with routine practice\u2014had access to patient histories and prior mammograms, while the AI only processed the most recent anonymized data with no added details.<\/span><\/p>\n

“Despite working from these X-ray images alone, the model surpassed individual experts in accuracy identifying breast cancer,” the blog said.<\/span><\/p>\n

Before AI can start saving lives, though, there are more hoopers to jump through, including additional research, clinical studies, and regulatory approval.<\/span><\/p>\n

“This work is the latest strand of our research looking into detection and diagnosis of breast cancer,” Shetty and Tse said. “Not just within the scope of radiology, but also pathology.”<\/span><\/p>\n

In 2017, Google published early findings regarding accurate detection of metastatic breast cancer from lymph node specimens. A year later, it introduced an algorithm<\/a> that can help doctors spot breast cancer more quickly and accurately in pathology slides.<\/span><\/p>\n

More on Geek.com:<\/span><\/i><\/p>\n