Researchers at Stanford University have recently utilized a very powerful type of computer learning algorithm called neural networks (a.k.a. deep learning algorithms) to classify skin lesions as cancerous or benign. The researchers used over 120,000 clinical images of skin disease to train the algorithm to classify the lesions. They claim it can classify lesions in photographic images similar to those taken with a mobile phone as accurately as a board-certified dermatologist. The researchers suggest this technique has potential to be used outside the clinic as a visual screen for cancer.

Click here to review the article in the journal Nature.