Investigators

Summary

Melanoma is a deadly cancer when caught late yet is a surgically curable disease when detected and treated at the earliest stages. Fortunately, most melanomas begin on the skin, and thus we have the opportunity to detect them using visual examination. However, early detection of melanoma can be challenging even for the most experienced dermatologists. This can lead to failure to detect melanoma when it is most treatable or to a high rate of removal of benign lesions. Both increase health care costs and put patients at risk. My research focuses on new techniques to improve the early detection of cutaneous melanoma. Computer vision technology can be applied to the analysis of specialized images of pigmented skin lesions taken using a handheld instrument called a dermatoscope. We are studying ways to use the objective, analytical power of the computer to take a dermatoscopic image of a pigmented skin lesion and rapidly quantify features such as symmetry, size, and color distribution. The system can then compare a lesion that has not been biopsied to a database of images from lesions for which a diagnosis has been made using histopathology following removal (the gold standard for diagnosis of melanoma and other skin lesions) to aid the user in making a decision of whether or not to biopsy. In addition, I am interested in understanding which patients benefit most from screening for melanoma and have several publications that examine melanoma epidemiology, detection patterns, and associations between melanoma screening and outcomes.