High Performed Skin Lesion Segmentation Based on Modified Active Contour

Skin lesions would be diagnosed as one of the most health - threaten issues in the era of cancer. Computer-aided diagnostic (CAD) systems use, analyze, and extract lesion traits in order to help physicians in the process of classifying its type and predicting severity rate. In this paper, a modified active contour (AC) segmentation has been proposed based on a cosine objective function. This modification has been manipulated with respect to snake force, curvature, and an appropriate Heaviside function. Experimental findings show that the proposed method indicates a segmentation accuracy of 97.13%, a sensitivity of 67.8%, specificity of 99.1%, dice coefficient (DC) of 66.3%, and Jaccard coefficient (JAC) of 50.24%.