Texture analysis for skin cancer diagnosis using dermoscopic images

Abstract

This paper provides a foundation to examine the dermoscopic images for skin cancer diagnosis. A dermoscopic image will often include textured areas that make up a major amount of the image. It is conceivable to organize and categorize such textures according to whether they are related with artifacts or if they reflect biological structure. Given the connection between structure, disease, and texture, it seems likely that quantitative measurements of texture might make it possible to characterize the tissues included inside a dermoscopic image. It has been shown that texture is a valuable characteristic for the characterization of skin cancer in dermoscopic images. The proposed system is comprised of two stages: the first is the extraction of information or features from dermoscopic images, and the second is the categorization of those images using a decision tree classifier. Based on the findings, it is possible to draw the conclusion that the extracted features have kept all of the information presents in the dermoscopic image that provides an overall accuracy of 98.89%