Industry 4.0 in the Health Sector: System for Melanoma Detection

Abstract

One of the most common cancers in humans is skin cancer [1], classified into two large groups: non-melanoma and melanoma [2, 3]. The latter is the most lethal type of cancer, as it ranks third in mortality in Mexico with 7.9% and represents 75% of the causes of death from skin cancer in the country. A study conducted between 2014 and 2018 in Mexico yielded a total of 3973 patients who died from melanoma [2]. Also, according to the American Society of Clinical Oncology ASCO and [4], early diagnosis is essential to combat this type of cancer. There are techniques that are used to make this diagnosis, such as visual inspection which is a non-invasive technique and invasive techniques such as biopsy, which help determine whether a skin lesion is benign or malignant [5]. In the present work, we propose the development of a dermatoscopic image classification model focused on melanoma detection based on deep learning. This artificial intelligence methodology is chosen as it has been shown to be robust and with high degrees of accuracy in image classification in any context and in particular in medical images. As will be observed below, the proposed model achieves an accuracy close to 90% with test images with an area under the receiver operating characteristic (ROC) curve of 0.95, which demonstrates a high performance in the classification task of the constructed model.