AI-Driven Early Detection and Classification of Skin Diseases Using Deep Learning and Mobile Edge Computing
TL;DR
An AI-based framework for automatic detection of skin disorders via deep learning and edge computing technology that makes the model more transparent through region of interest identification in clinical images, whereas the lightweight web application makes predictions along with disease prediction confidence and visualization of diagnosis.
Abstract
Skin disorders are experienced by millions of people across the globe, thus, calling for timely and accurate diagnosis in order to ensure proper treatment and good prognosis of the disease. Nevertheless, lack of access to dermatologists in some parts of the world, especially in rural areas and poor regions, often hinders proper and timely diagnosis of the condition. This work proposes an AI-based framework for automatic detection of skin disorders via deep learning and edge computing technology. Grad-CAM powered explainable artificial intelligence makes the model more transparent through region of interest identification in clinical images, whereas the lightweight web application makes predictions along with disease prediction confidence and visualization of diagnosis. Evaluation on HAM10000 and ISIC datasets attained 94.8% accuracy, 93.6% precision, 94.1% recall, 93.8% F1-score, and 95.4% mAP. The proposed solution presents an effective and scalable solution for AI-assisted dermatological diagnosis.