Cancer treatment may have just taken a big leap into the computer age.
Stanford Univeristy reasearchers are reporting significant success in training computers to anaylze microscope slide of breast cancer biopsies, with keener eye than any human pathologist.
Since the early 20th century, pathologists have been squinting into the microscopes, looking for a handful of features in biopsies tumers samples, that enable them to classify how agreesive cancer is.
That information helps doctors decide how to treat patient.
Today, armed with sophisticated software, powerful computers get quiet skill at pattern recognition, identifying faces for example.
The Stanford researchers thought computers may be able to learn to evaluate cancer by biopsies, too.
To do that, Dfini cooler and colleages study with a set of biopsies slide and use to train pathologist. The slide was scanned into the computer which meansured not just a handful of features human pathologist might review, but thousands of characteristics on each image.
"And we plugged it into a machine learning algorithem that looked at survival data and try to figure out which of these features were good feature in term of survival, which were bad feature, and which were not relevant at all