Analysis Of Support Vector Machine And Resnet-50-Based Skin Cancer Detection

Abstract

This paper focuses on the application of algorithms available in machine learning, particularly Support Vector Machines (SVM) as well as Resnet-50, in the classification of dangerous skin cancer from epiluminescence microscopy images. The study analyzes effectiveness of the two models on accuracy evaluation metrics that include confusion matrix, graphical plots, Receiver Operating Characteristics (ROC) and attempts to find which model more optimally detects skin cancer. Previous studies indicate that Resnet-50 outperforms SVM in detection accuracy capabilities. Thus, the purpose of this paper is to also showcase the ability to enhance perception accuracy for skin cancer by integrating both models. The results of this study are clinically relevant. With the implementation of computer-aided diagnosis (CAD) systems, clinicians are now able to make reliable diagnoses of skin cancer which lessens the degree of subjective variability between different clinicians and enhances clinical objectivity. The study emphasizes the diagnosis and treatment of skin cancer using machine learning models, which improves patient consequences. The abstract highlights significant insights on the performance of models present in machine learning for detecting skin cancers and becomes a useful resource for clinicians and researchers who consider adapting the use of machine learning in skin cancer detection.