Skin cancer meets its worst nightmare: IBM

Skin cancer may have found its worst enemy: IBM’s cognitive computers.

VentureBeat has learned that IBM Research will announce a partnership today with New York’s Memorial Sloan Kettering Cancer Center on technology that, in tests, successfully evaluated patterns in medical images and detected even the most deadly forms of skin cancer as much as 97 percent of the time.

The project, run out of the Multimedia Analytics group at IBM Research’s Yorktown Heights, N.Y. headquarters, is Big Blue’s answer to the U.S. Surgeon General’s 2014 “Call to Action” to defeat skin cancer. Citing its crippling massive annual national impact — 9,000 deaths and $8 billion in health-care costs — IBM and Memorial Sloan Kettering set out to find a way to apply automated analytics to better detect the disease.

Led by IBM researcher Noel Codella, the project builds on the company’s work on machine learning techniques aimed at automatically picking out things like dogs or cats in imagery, or even picking out a dog or a cat in a photo.

At the same time, Codella told VentureBeat, Memorial Sloan Kettering has been building a database of dermatological images that show different kinds of skin lesions and other elements of diseases and tie them to specific clinical properties.

“Our cognitive systems are using this data to learn what types of features and patterns are most frequent in melanoma to help recognize the disease in images,” Codella said in a phone interview.

Codella added that IBM’s technology has proven adept at analyzing large numbers of images far quicker and with a more finely detailed level of measurement than any doctor could do. The system is designed to evaluate an image in less than a second.

While Codella’s group often works with IBM’s Watson division, Watson — which famously beat the world’s top Jeopardy players and is now being used to tackle large-scale computing problems in everything from finance to health — was not involved in the skin cancer project.

IBM used its technology in controlled tests of more than 3,000 cases of melanoma and other skin lesions. Codella explained that the system was able to identify positive and negative cases of skin cancer with 95-plus percent accuracy.

The highest level of accuracy humans have achieved is 84 percent, Codella said, while some automated approaches have hit 90 percent.

For now, this technology is still in the research stage, and Codella said it could be some time before it is available to dermatologists everywhere. That means, of course, that it could fail to emerge from IBM’s research labs. But if it eventually emerges as a tool available to doctors, skin cancer patients could have their best ally ever in combating the disease.