Melanoma—the most serious form of skin cancer—varies in size, shape and severity, and although pathologists and researchers often intuitively differentiate between types of melanomas, the variations have never been formally quantified and documented.
A project involving RENCI and researchers in the University of North Carolina School of Medicine and the departments of computer science and statistics and operations research aims to change that by using image analysis techniques to improve melanoma classification. The research team, led by Nancy E. Thomas, a professor and M.D. in the School of Medicine's dermatology department, examines images made from biopsy slides of both cancerous and healthy skin cells.
A team that includes RENCI Senior Visualization Researcher David Borland, Marc Niethammer, assistant professor in the UNC computer science department, and Steve Marron, a professor in UNC's statistics and operations research department, uses the image data to train "smart" algorithms to identify physical characteristics of cells in the images, such as whether they are cancerous or benign, and the size, shape and color of healthy tissue and tumors.
The information that is extracted from the data can then be used to classify melanoma variations. These details also help the researchers develop evidence-based models of melanoma cell descriptions that in time could help doctors determine the seriousness of melanoma cases more quickly and with better accuracy.