Google uses machine learning to detect breast cancer better than pathologists
Google Inc. is taking the good fight to battling cancer, revealing that its machine learning platform has delivered better breast cancer diagnosis rates than pathologists.
Detailing the results in a new research paper, Google explained how it applied machine learning, predictive analytics and pattern recognition to achieve an 89 percent accuracy rate. That’s quite a bit ahead of an average score of 73 percent for a human pathologist looking at biological tissue samples on a slide.
“Pathologists are responsible for reviewing all the biological tissues visible on a slide. However, there can be many slides per patient, each of which is 10+ gigapixels when digitized at 40X magnification,” the Google Research Blog explained Friday. “Imagine having to go through a thousand 10-megapixel photos, and having to be responsible for every pixel. Needless to say, this is a lot of data to cover, and often time is limited.”
The problem was approached using deep learning, a branch of machine learning that uses neural networks simulated in software to help machines learn without being explicitly programmed. In this case, the system used images from the Radboud University Medical Center to train algorithms optimized for localization of breast cancer that had metastasized to lymph nodes adjacent to the breast.
The results of the test weren’t perfect, described as at times delivering “busy” heat maps. But the results were nonetheless impressive given they were quickly generated using Google’s artificial intelligence platform, whereas pathologist were given an unlimited amount of time to come to their own conclusions.
The images on the left above are from two lymph node biopsies. In the middle are earlier results of Google’s deep learning tumor detection, and on the right are current results. The researchers point to “visibly reduced noise (potential false positives)” between the two versions.
The researchers said the technology is not yet close to replacing pathologists, as the white noted that while the algorithms do a great job for the tasks for which they are trained, they lack the breadth of knowledge an experienced pathologist would have.
The authors of the paper may have been somewhat modest in predicting how fast machine learning will replace humans in medicine, but others are not so restrained. Ziad Obermeyer and Ezekiel J. Emanuel argued in a paper titled “Predicting the Future — Big Data, Machine Learning, and Clinical Medicine” submitted to the New England Journal of Medicine in 2016 that not only will the use of machine learning more reliably predict outcomes and improve prognosis, but it will also displace much of the work of pathologists as well.
Since you’re here …
Show your support for our mission with our one-click subscription to our YouTube channel (below). The more subscribers we have, the more YouTube will suggest relevant enterprise and emerging technology content to you. Thanks!
Support our mission: >>>>>> SUBSCRIBE NOW >>>>>> to our YouTube channel.
… We’d also like to tell you about our mission and how you can help us fulfill it. SiliconANGLE Media Inc.’s business model is based on the intrinsic value of the content, not advertising. Unlike many online publications, we don’t have a paywall or run banner advertising, because we want to keep our journalism open, without influence or the need to chase traffic.The journalism, reporting and commentary on SiliconANGLE — along with live, unscripted video from our Silicon Valley studio and globe-trotting video teams at theCUBE — take a lot of hard work, time and money. Keeping the quality high requires the support of sponsors who are aligned with our vision of ad-free journalism content.