Analysis and Classification of Skin Spot Images by Extraction of Texture Attributes for Identification of Melanoma

Marcos AM Almeida
2021 Biomedical Journal of Scientific & Technical Research  
The use of Artificial Intelligence techniques has increasingly motivated scientists to detect and classify medical images. It is no different in the field of dermatology, with the use of machine learning to track anomalies of the skin and diagnose cancer, using data sets consisting of dermatoscopic images. The texture analysis of medical images is the result of digital image processing able to extract representative features which may have elements that are impossible to detect through human
more » ... ion. This research reports on diagnostic prediction results from a set of images of human tissues with nevus and melanoma characteristics as represented by appropriate machine learning tools in order to assist physicians specialized in prognoses undetectable by the human eye. The results were 96.67% accurate against the benchmark. The study database was classified using the Gradient Boosting classifier, a process that extracts skin images features through the red, green and blue channels.
doi:10.26717/bjstr.2021.35.005676 fatcat:2daasooe4jhj5p7al7aem4azly