First-order Feature Extraction Methods for Image Texture and Melanoma Skin Cancer Detection
Masna Wati, Haviluddin Haviluddin, Novianti Puspitasari, Edy Budiman, Robbi Rahim · Journal of Physics Conference Series · 2019
Skin cancer is a disease characterized by the growth of uncontrolled skin cells, which can damage surrounding tissue and spread to other body parts. The purpose of this study was to facilitate early recognition of skin cancer by applying the first-order extraction method using 6 parameters i.e. contrast, variance, standard deviation, kurtosis, mean and smoothness, for feature extraction based on texture to obtain a good level of accuracy and classification methods using Multilayer Perceptron Neural Network (MLP NN). The results of diagnostic identification consist of 2 outputs, i.e. melanoma and not melanoma. From the research, accuracy measurements were obtained through 4 sets of test images using melanoma and non-melanoma images and the results showed that the lowest level of accuracy was 81.81% and the highest level of accuracy was 85.71% so that the overall accuracy rate is 83.86%.