Deep Learning Skin Disease Diagnosis and Prognosis Based on Artificial Intelligence

Abstract

One of the most common disorder spread between people is dermatology, which have heavily touched to peoples live, these diseases can result from various factors (bacteria, infection or radiation), Identifying these diseases in the initial phase ensure improvement in healing likelihood. In this research, an artificial intelligence system represented by deep learning is used, the model built based on an architecture of Convolutional Neural Network (CNN) along with Visual Geometry Group (VGG16) network to detect three kinds of diseases, identified by “nevus, melanoma and seborrheic keratosis”. A total of 1,403 dataset sourced from Kaggle were used for training and testing. An accurate result of 99.31% were gained, in order to estimate the performance of the methodology suggested. These findings revealed the robustness of CNN-based system to classify the dataset in high accuracy. The presented model main objective is to distinguish between unusual kind of skin disease categories, employing several performance valuations, involving (accuracy, precision, f1-score, recall, and support, and highlighting on most related methodologies in this field.