IBM detects skin cancer more quickly with visual machine learning

IBM detects skin cancer more quickly with visual machine learning

Cognitive computing could improve the ability to spot melanoma early on

Skin cancer can be detected more quickly and accurately by using cognitive computing-based visual analytics, researchers at IBM Research have found, in collaboration with New York's Memorial Sloan Kettering Cancer Center.

In a scan of 3,000 images, IBM technology was able to spot melanoma with an accuracy of about 95 percent, much better than the 75 percent to 84 percent average of today's largely manual methods.

"The technology can pull on massive amounts of data to help the doctor make more informed decisions," said Noel Codella, a multimedia analytics researcher in the cognitive computing group at the IBM T. J. Watson Research Center in Yorktown Heights, New York.

Once the technology is commercialized, the cognitive computing approach will be able to scan the images in less than a second, much more quickly than humans can.

Such work could go a long way to more effectively treating skin cancer, which afflicts nearly 5 million[m] people a year in the U.S. alone, according to the U.S. surgeon general.

Cognitive computing could bring a new efficiency to recognizing melanoma. Machine learning algorithms could continually improve the ability of the system to identify the disease. Over time, this approach might even be able to spot cases that would be too difficult to pinpoint by a doctor.

As with any form of cancer, skin cancer is best diagnosed early; a cognitive computing detection system could flag potential trouble areas before they become noticeable by humans.

IBM's approach involves the use of multiple tests. One set of tests could look for unusual color distributions or texture patterns on the skin. Another series of tests could also identify the rapid progression of lesions, or deviations from normal growth, compared to the rest of the body or to other people with similar genetic or demographic characteristics

The system would weigh the results of each test. "We don't take a single approach. So we study a variety of approaches, see how they work, and see if they can be combined in some way so they work better," Codella said.

This work builds off research on machine learning technologies that aid computers in recognizing objects within images. It leverages a number of the company's advanced technologies, including a visually oriented machine learning architecture called the IBM Multimedia and Analytics System.

It also uses an IBM system for analyzing medical images, called Medical Sieve, and a visual recognition and search system called Intelligent Video Analytics.

IBM Research will continue to work with Sloan Kettering to develop additional measurements and approaches to further refine diagnosis, as well as refine their approach through larger sets of data.

Joab Jackson covers enterprise software and general technology breaking news for The IDG News Service. Follow Joab on Twitter at @Joab_Jackson. Joab's e-mail address is

Follow Us

Join the New Zealand Reseller News newsletter!

Error: Please check your email address.

Tags popular scienceapplicationsIBMdata miningsoftwarehealth careindustry verticals


Show Comments