Black Widow-Based Optimal SVM and Fuzzy Segmentation for Classifying Skin Diseases

Abstract

The primary human organ and external barrier is the skin, which is composed of seven layers. Skin cancer is the fourth most common cause of non-fatal disease risk, according to the World Health Organisation (WHO). Classifying skin diseases is a difficult problem in the medical field because of overfitting, erroneous results, higher computational costs, and other factors. We introduced a brand-new method for classifying skin diseases called support vector machine–based black widow optimisation, or SVM-BWO. Images of five distinct skin diseases—psoriasis, paederus, herpes, melanoma, and benign—as well as healthy ones are selected for this project. The original input images' noise is eliminated through the preprocessing step. The skin lesion region is then segmented using the new fuzzy set segmentation algorithm. The colour, gray-level co-occurrence matrix texture, and shape features are then taken out for additional processing. The SVM-BWO algorithm is used to classify skin diseases. Since MATLAB-2018a is used for implementation tasks, ISIC-2018 datasets were used to gather the dataset images. Various types of performance analyses are conducted experimentally using cutting-edge methods. In any case, the suggested methodology performs better than alternative approaches, with a 92% classification accuracy.