Google Uses AI System To Detect Cancer

Posted: Mar 6 2017, 2:05am CST | by , Updated: Mar 6 2017, 2:15am CST, in News | Latest Science News

 
Google Uses AI System to Detect Cancer
Credit: Google Research Blog

Google's deep learning algorithm is a very sensitive tool for identifying breast cancer

Detecting cancer sooner rather than later can help prevent worst outcomes in many cases. Yet diagnosing different types of cancers in their earliest stages is not often possible. It requires years of training to gain the expertise and experience to do well in detecting cancerous tumor.

Even with extensive training, some symptoms may be hard to distinguish from the signs of other similar diseases. For example, agreement in diagnosis for some forms of breast cancer can be as low as 48% and this percentage of disagreement is not surprising given the massive amount of information needed to review in order to make an accurate diagnoses and often within a limited time.

To overcome these issues, Google researchers have developed an automated detection algorithm that is effective at identifying breast cancer. This artificial intelligence program can go through thousands of slides of cancer cells and can recognize specific patterns in huge dataset.

Though the technology is young and it must prove itself in many more tests, Google researchers believe that it is possible to train an algorithm that either match or exceed the performance of a pathologist who has extensive time to examine laboratory samples and to make a diagnosis. Currently, algorithm reaches the score of 89% which is significantly higher (73%) than that of a pathologist with no time constraint.

“We were not the only ones to see promising results, as other groups were getting as high as 81% with the same dataset. Even more exciting for us was that our model generalized very well, even to images that were acquired from a different hospital using different scanners.” Google statement reads.

Like many other deep learning algorithms, Google AI is far from being the perfect. It performed well for the task for which it was trained, but lacked certainty in detecting abnormalities that it was not explicitly trained to classify. To ensure the best outcome, the algorithm needs to be incorporated in a way that complements the pathologist’s workflow.

“We envision that algorithm such as ours could improve the efficiency and consistency of pathologists.” Google researchers wrote in the blog.

“These algorithms could enable pathologists to easily and accurately measure tumor size, a factor that is associated with prognosis.”

This story may contain affiliate links.

Find rare products online! Get the free Tracker App now.


Download the free Tracker app now to get in-stock alerts on Pomsies, Oculus Go, SNES Classic and more.

Latest News

Comments

The Author


Hira Bashir covers daily affairs around the world.

 

 

Advertisement

comments powered by Disqus