A fuzzy-based histogram analysis technique for skin lesion classification in dermatology clinical images

“Melanoma is a form of a skin cancer that can be deadly if not treated in the early stages. Imaging of skin lesions provides an important diagnostic aid to dermatologists in the detection of malignant melanoma. There are several important features used for detecting melanoma, including lesion asymmetry, border irregularity, color variegation and skin lesion diameter. It is believed that color plays an important role in the diagnosis of malignant melanoma. There are “melanoma colors” lying in the color space. Color variegation usually involves examining skin lesions for certain colors or groups of colors that are indicative of melanoma. An alternative approach to melanoma detection is to identify colors that are most characteristic of benign lesions and to evaluate skin lesion color deviation with those characteristic colors. Accordingly, a perception-based framework provides the capability to make quantitative assessments related to the degree of similarity of skin lesions to benign lesion standards. In this research, a fuzzy logic- based approach for skin lesion color analysis is investigated for differentiating benign skin lesions from melanomas in dermatology clinical images. Experimental results are reported over a data set of 258 clinical images, including 129 benign lesions and 129 melanomas. Two different membership functions, stepwise and trapezoidal, are used in order to represent the fuzzy sets. A fusion approach is presented of the trapezoidal membership function-based approach with an existing skin lesion percent melanoma color feature for enhanced skin lesion classification”--Abstract, page iii