Abstract The incidence of skin cancer, one of the most rapidly growing and common malignancies, has been rising at a faster pace worldwide. This cancer can be disfiguring or even deadly if not recognized in the earlier stages, and cancer identification has long been considered a critical concern. Computer-dependent analysis is the latest clinical tool, allowing skin cancers to be reliably analyzed within a minimum focus period. This chapter describes novel image-processing approaches for skin cancer classification using dermoscopy images. The reliability of a particular skin lesion classification method depends on the correct selection of features, object extraction strategy, and the system’s process classifier. The motive of this research is to investigate and select an algorithm for skin cancer classification that can accurately group the lesions as malignant or benign. Initially, recognition of malignant and benign skin lesions by means of GLCM, texture, and wavelet-based features is proposed, using support vector machine, Naive Bayes, and artificial neural network classifiers. Then the competence of these classifiers is analyzed based on precision, sensitivity, and specificity to determine the most effective classification technique.
Paper
Full text
Analysis of skin lesions using machine learning techniques
Semantic Scholar · Computer Science · 2020
Abstract
Abstract The incidence of skin cancer, one of the most rapidly growing and common malignancies, has been rising at a faster pace worldwide. This cancer can be disfiguring or even deadly if not recognized in the earlier stages, and cancer identification has long been considered a critical concern. Computer-dependent analysis is the latest clinical tool, allowing skin cancers to be reliably analyzed within a minimum focus period. This chapter describes novel image-processing approaches for skin cancer classification using dermoscopy images. The reliability of a particular skin lesion classification method depends on the correct selection of features, object extraction strategy, and the system’s process classifier. The motive of this research is to investigate and select an algorithm for skin cancer classification that can accurately group the lesions as malignant or benign. Initially, recognition of malignant and benign skin lesions by means of GLCM, texture, and wavelet-based features is proposed, using support vector machine, Naive Bayes, and artificial neural network classifiers. Then the competence of these classifiers is analyzed based on precision, sensitivity, and specificity to determine the most effective classification technique.