Google trains computers to detect breast cancer
07 March 2017
Google is harnessing the power of computer-based reasoning to detect breast cancer, training the tool to look for cell patterns in slides of tissue, in much the same way as the brain of a doctor.
According to new findings, the approach enlisting machine learning, predictive analytics and pattern recognition had achieved 89 per cent accuracy, beyond the 73 per cent score of a human pathologist.
''We showed that it was possible to train a model that either matched or exceeded the performance of a pathologist who had unlimited time to examine the slides,'' according to a blog item in Google Research posted by technical lead Martin Stumpe and product manager Lily Peng.
To reach an accurate diagnosis, pathologists had to sift through massive amounts of data, including slides containing cells from tissue biopsies, thinly sliced and stained that had to be scanned to detect abnormal cells. And time was a critical factor.
There could be many slides per patient, each containing over 10 gigapixels when digitised at 40 times magnification, according to the Google team.
''Imagine having to go through a thousand 10-megapixel photos, and having to be responsible for every pixel,'' the team wrote.
Google used a kind of artificial intelligence called deep learning to analyse thousands of slides of cancer cells provided by a Dutch university. In Deep learning computers are taught to recognise patterns in huge data sets, which was very useful for visual tasks, such as looking at a breast cancer biopsy.
With 230,000 new cases of breast cancer every year in the US, Google hoped its technology would help pathologists better treat patients. The technology was not designed to or capable of, replacing human doctors.
"What we've trained is just a little sliver of software that helps with one part of a very complex series of tasks," said Peng CNN Money reported. "There will hopefully be more and more of these tools that help doctors (who) have to go through an enormous amount of information all the time."