AI beats doctors at identifying skin cancers
ARTIFICIAL intelligence is better than doctors at diagnosing some cancers, research has shown.
The international study involved machines that were trained to detect signs of skin cancer before being tested against 58 dermatologists.
Yesterday the Health Secretary, Jeremy Hunt, said extra funds for the NHS must be used to expand the use of artificial intelligence to diagnose patients.
The computer network was taught by being shown 100,000 images of malignant melanomas and benign moles marked with a diagnosis. It was then pitted against senior doctors to diagnose 100 of the most difficult lesions.
The machines correctly diagnosed the malignant cases in 95 per cent of cases – significantly more than the 87 per cent accuracy achieved by dermatologists, the study found.
The artificial intelligence also misdiagnosed fewer benign moles as malignant melanoma, the most deadly type of skin cancer, meaning fewer patients would endure needless surgery.
Researchers from Germany, the United States and France used a form of artificial intelligence known as a deep learning convolutional neural network (CNN). This is an artificial neural network inspired by the biological processes at work when nerve cells in the brain are connected to each other and respond to what the eye sees.
The CNN learns from images that it “sees” and teaches itself to improve its performance in a process called machine learning.
Prof Holger Haenssle, the lead author, from the University of Heidelberg, Germany, said: “These findings show that deep learning convolutional neural networks are capable of out-performing dermatologists, including extensively trained experts, in the task of detecting melanomas.” The study was published in the Annals of Oncology.