Nail Disease Prediction using a Deep Learning Integrated Framework

Abstract

Deformities that may develop in the nail unit are referred to as "nail diseases". Even though the nail unit is regarded as a skin accessory, these diseases have their own distinct set of signs, causes, and outcomes that may or may not be connected to other illnesses. It can be challenging to identify and recognize certain nail disorders, hence this research suggests a novel deep learning framework that can identify and categorize these diseases from images. The framework is specifically made to identify nine different types of nail diseases based on nail colour, including black, blue, grey, purple, red, white, and yellow coloured nails with beau's lines, hyperpigmentation, onychomycosis, subungual hematoma, yellow nail syndrome, psoriasis, koilonychias, paronychia, pincer nails, leukonychia, and onychorrhexis. In order to do this, feature extraction is carried out using a mix of convolutional neural networks (CNNs). The researchers created their own dataset to assess the effectiveness of their suggested framework because there isn't a comprehensive dataset readily accessible for this assignment. The outcomes were contrasted with those of other cutting-edge algorithms, including RF, SVM, ANN, and KNN, which are renowned for excelling in feature extraction. With an accuracy of 87.33%, the suggested framework distinguished between varieties of nail conditions with comparable performance.