‘Automated dermatologist’ detects skin cancer nearly as well as doctors

This is an archived article and the information in the article may be outdated. Please look at the time stamp on the story to see when it was last updated.

STANFORD, Calif. — If that rash, bump or spot on your skin is causing you to fret, you soon might be able to gauge the seriousness of the abnormality without stepping foot in a doctor’s office.

Considering the fact 5.4 million new cases of skin cancer are diagnosed annually in the United States, early detection of the disease is key to boosting survival rate and even gut feelings should be a good reason to get yourself checked.

The Artificial Intelligence Algorithm can identify skin cancer nearly as well as a professional doctor. Image recognition technology was used in the study and created by researchers at Stanford University.

The technology involved in the study analyzed 130,000 pictures of moles, rashes and lesions to come up with conclusions regarding the abnormality based solely on the images.

The diagnosis created by the algorithm was compared with the opinion of 21 human dermatologists. According to the study, creators claim the technology performed with at least 91 percent accuracy.

When determining the seriousness of a skin abnormality, dermatologists first typically rely on a visual examination.

An existing “deep learning” algorithm built by Google for image classification was used as the base for the project.

Creators then added the thousands of pictures and names of cancerous and benign bumps, lesions and rashes to be examined.

In the future, researchers suggest the algorithm could be used to create a mobile app for spotting skin cancer at home.