UNITED NEWS INTERNATIONAL (UNI) — New research suggests artificial intelligence may be better than doctors at detecting skin cancer.
A study published May 29 in the journal Annals of Oncology included researchers from Germany, the U.S. and France.
These researchers programmed an algorithm known as a deep learning convolutional neural network — or CNN for short.
They showed the CNN thousands of images of skin lesions and trained it to recognize the differences between dangerous and benign cases.
The machine then taught itself how to get better and better at correctly diagnosing malignant melanomas.
Researchers said the CNN correctly detected 95 percent of melanomas while the doctors caught an average of 86.6 percent.
Even though the doctors’ scores improved after they were given more information about individual patients, the CNN program still performed better.
And it only used pictures.
The technology could one day be used in skin cancer screening and possibly help save lives.