The Impact of Segmentation on the Accuracy and Sensitivity of a Melanoma Classifier Based on Skin Lesion Images

In the United States alone, there were an estimated 76,380 new cases of melanoma and an estimated 6,750 deaths due to melanoma in 2016 (Siegel, Miller, & Jemal, 2016). Early screening can increase life expectancy (Freedberg et al., 1999), but melanoma left undiagnosed can be fatal. Dermatologists use many heuristic classification methods to diagnose melanoma (Argenziano et al., 1998; Nachbar et al., 1994), but to limited success with only 65 80% accuracy (Argenziano & Soyer, 2001). A tool capable of aiding physicians to classify skin lesions could potentially save numerous lives each year.