Automated Melanoma Skin Disease Detection Using the Spike Herd Search Algorithm

Abstract

Melanoma is rare and can shorten a patient's life. It can be cured if found early. Early cost, pain and timing prediction affect patients. Thus, many researchers use Artificial Intelligence (AI) to identify Melanoma skin cancer. However, complicated skin lesions, illumination, and a lot of training data make diagonalising Melanoma skin cancer diagnosis tough. The Spiking Herd Search Learning (SHSL) approach for optimal feature selection-based classification promotes the category to overcome this issue. The method combines a unique hybrid metaheuristic algorithm, Horse herd Optimisation Algorithm (HOA), and Squirrel Search Optimisation (SSO) with deep hybrid learning as Spiking Neural Network (SNN) and Convolutional Neural Network (CNN) classifiers. In preprocessing, a Gaussian filter removes skin lesion noise using Image Processing techniques. The k-means clustering method performs the segmentation and excellent feature extraction techniques locate the lesion exactly. The Horse herd Optimisation Algorithm (HOA) and Squirrel Search Optimisation (SSO) algorithms and techniques are used toimprove Spike Neural Network classification (SNN) accurate features. Finally, using the DermIS Image dataset, the SHSL algorithm is evaluated and compared to other methods. The proposed method detects Melanoma Skin Disease with 98.91% accuracy.