NailScan - Nail Disease Classification using Deep Learning and Treatment Recommendation System

Abstract

"In the medical field, early detection and treatment of diseases are crucial for successful outcomes. Various parts of the body, including nails, can provide important indicators of underlying health issues. Changes in nail color or shape may signal dysfunction or disease, with untreated conditions potentially spreading to other nails or causing infections. Additionally, neglected nail diseases can impede nail growth and lead to further infections in other areas of the body typically covered by nails. Diagnosing nail diseases is often done by looking at them, which takes a lot of time. Experts struggle to tell them apart because they look so similar. And in rural areas, it's hard for people to see a doctor. Due to these factors the author has come up with a solution using Deep Learning which classifies the nail disease when an image of the nail disease is entered to the system. The author has combined publicly available datasets to get the final dataset which consist of eight nail diseases. This is more than what has already been done in previous nail disease classification systems. An ensemble model was implemented to classify the nail disease using two Convolutional Neural Network models. Additionally, the system provides treatment recommendations to the user using the OpenAI API. For the implementation results, an accuracy of 86.86% and a loss of 0.56 was obtained for the ensemble model. The confusion matrices and the classification report provides the evaluation of the classification."