IBM Research partners to find melanoma via cognitive computing

IBM Research

IBM Research in Australia is partnering with Melanoma Institute Australia on identifying melanoma using cognitive technology. The team will conduct research to build on IBM's work with MoleMap. MoleMap uses visual analytics to analyze more than 40,000 data sets that include images and text.

In its research, IBM Research will analyze images of skin lesions and look for clinical patterns in early stage melanoma. According to the announcement, the aim is to reduce unnecessary biopsies and better understand skin cancer.

IBM Research hopes to improve the accuracy of its machine learning algorithms through the analysis of de-identified data, which include over 1 million images from 9,000 Australian and New Zealand patients, in addition to clinical notes, the announcement explained. The tech aims to understand skin cancers like melanoma, basal cell carcinoma and squamous cell carcinoma as accurately through lower resolution clinical images, compared with dermoscopy images.

WEBINAR

Webinar: Meet the Challenge of Complex Protein Expression

As market demand continues to rise for more potent and effective therapeutics, biologic pipelines are evolving from standard antibody formats to next-generation biologics (NGBs). In this webinar we will discuss and demonstrate application through case studies, two significant enhancements to Lonza’s GS Xceed® expression system to help address the challenges of NGBs.

"Research that enables the earlier detection of melanoma is likely to save more lives in the future. The five-year survival rate for melanoma is only 64 percent once the disease reaches the lymph nodes. However this rises to 95 percent if detected before then," said Graham Mann, research director at Melanoma Institute Australia, in the announcement. "Diagnosing melanoma with the naked eye is only about 60 percent accurate, but dermoscopy can lift that to over 80 percent. Research using automated analysis of images could provide the next gain in accuracy, especially where dermoscopy is hard to access."

Melanoma Institute Australia has the largest melanoma research database in the world and operates the world's largest melanoma research and treatment facility.

In 2015, IBM Research worked with MoleMap to complete a retrospective analysis of 40,000 images, dermatology opinions and diagnosis on each image across three types of skin cancer and 12 benign disease groups. "This formative testing on historical data suggested an accurate detection of melanoma from 12 various benign skin diseases of 91 percent on dermoscopy images, and 83 percent on clinical photography images," the announcement explained.

Since 1997, MoleMap has addressed more than 6 million lesions for over 200,000 patients across 50 clinics in the U.S., New Zealand and Australia.

The research done between IBM Research and Melanoma Institute Australia will test this performance. "Melanoma Institute Australia and MoleMap will help IBM Research to further train and validate the algorithms to help improve the understanding and identification of early stages of melanoma, which could help to improve patient care," the announcement said.

Joanna Batstone, vice president and lab director at IBM Research, explained that cognitive computing is able to process large amounts of complex data, including images and text, more quickly than manual methods. In addition, machine learning tech improves as data is fed into it. "This initiative could inform future research and, potentially, the development of offerings that could have enormous implications for both the Australian public and the health system," Batstone said.

- here's the press release

Read more on

Suggested Articles

J&J’s Ethicon unit received an FDA clearance for its Vistaseal applicators that spray a biologic sealant from Grifols to help stem surgical bleeding.

Bio-Techne’s urine test has received a breakthrough device designation from the FDA for ruling out unnecessary tissue biopsies.

Qiagen launched a one-stop shop compiling publicly available genomic data, scientific literature and phenotypic information on potential superbugs.