Artificial intelligence can diagnose skin cancer as well as a trained doctor

Artificially intelligent computer systems are now able to recognise skin cancer just as well as human scientists, according to a new report. Researchers from Stanford University used the internet to gather 130,000 pictures of skin lesions, representing over 2,000 different diseases. They then used these pictures to train an artificial intelligence algorithm - originally developed by Google to differentiate between cats and dogs - to tell the difference between cancerous and non-cancerous lesions. After training the algorithm, they tested it on 370 high-quality, biopsy-confirmed images, and compared its responses with those of 21 professional dermatologists. In its diagnoses of the skin lesions, which represented the most common and deadliest skin cancers, the algorithm matched the performance of the dermatologists.