A Novel Segmentation Based Technique for Detection of Skin Lesion using Deep Learning

Abstract

Skin cancer is a common and serious disease in modern times. Early detection of skin cancer is crucial for reducing mortality rates. Due to the rapid increase in skin cancer cases, there is a sudden need to develop a Computer-Aided Diagnosis (CAD) framework that can accurately diagnose skin lesions. CAD models may aid doctors in identifying problems, resulting in improved diagnostic outcomes. Our analysis has shown that the Deep Neural Network exhibits higher accuracy compared to other machine learning methods included in the research. This paper presents a new method for categorizing skin lesions. The method includes pre-processing, bilateral filtering approaches, segmentation, ResNet-50 feature extraction, and feature selection utilizing the Whale Optimization Method (WOA) algorithm. The DBN model achieved the maximum accuracy of 96.1% in the suggested system. The suggested research aids in the early detection of seven types of skin cancer, allowing for validation and suitable treatment by medical professionals.