Sign In

Communications of the ACM

ACM TechNews

Google Computers Trained to Detect Cancer


Top view of a female breast with a tumor.

A new breast cancer detection technique developed by Google can train a computer to search for cell patterns in tissue slides.

Credit: Getty Images

Google has developed a new breast cancer detection technique that utilizes machine learning, predictive analytics, and pattern recognition to train a computer to search for cell patterns in tissue slides.

Google researchers found the system can autonomously learn to identify pathology by being fed billions of images. They say the system is 89% accurate, versus a pathologist's 73% score.

The team emphasizes their algorithms will supplement, and not replace, human physicians. They note, for example, a computer cannot spot abnormalities that it has not been trained to classify, such as different cancer types. However, a computer can automatically alert the physician to the most critical images, or it could help them more easily and accurately quantify tumor size.

"To ensure the best clinical outcome for patients, these algorithms need to be incorporated in a way that complements the pathologist's workflow," the researchers say.

From The Mercury News
View Full Article

 

Abstracts Copyright © 2017 Information Inc., Bethesda, Maryland, USA


 

No entries found