Skin Cancer Detecting Robot
August 7, 2018 | Terry Sharrer
Last year, Stanford researchers reported on an artificial intelligence technology called a “convoluted neural network” to screen images in assessing possible melanoma lesions. This piece confirms Stanford’s study with findings from dermatologists at the University of Heidelberg—that a robot is more accurate in spotting melanomas than a human pathologist. 86.6% correct for the pathologists; 95% correct for the robot. The robot trained on 100,000 images before the competition. MORE
Image Credit: Matt Young and Stanford News