Proposed optimized active contour based approach for accurately skin lesion segmentation

The presented work suggests a robust approach for skin lesion segmentation in the state of the art of optimized active contours and level sets according to the estimation of both restriction and energy forces. The proposed system would separate accurately the interest region by defining the appropriate contour or curvature. Accordingly, the following optimizers have been utilized and tested; particle swarm optimization (PSO), genetic algorithm (GA), grey wolf optimization (GWO), whale optimization algorithm (WOA), bee colony optimization (BCO), and interior-point optimization (IPO) for optimizing the suggested trigonometric regularization functions. Experimental findings have recommended the use of the proposed Sine-IPO algorithm which targeting the best-segmentation of different types of skin lesions in dermoscopy images. In addition, different performance metrics have been utilized; true positive (TP), true negative (TN), false positive (FP), and false-negative (FN) in order to guarantee the appropriate segmentation accuracy. The results depicted the superiority of using IPO based Sine function with accuracy (about 96.23%), sensitivity (about 66.48%), specificity (about 99.45%), dice coefficient (DC) (about 67.43%), and Jaccard coefficient (JAC) (about 53.63%).