Skin Disease Detection Using Neural Networks
Meenu Gupta, Rakesh Kumar, Nandan, Anil K. Pradhan, Ahmed J. Obaid
Abstract
Skin diseases like acne, psoriasis, eczema, and dermatitis affect millions worldwide. Skin cancer and melanoma are diseases that happen due to exposure to UV radiation. The early detection of skin diseases is crucial for effective treatment and prevention of further complications. This work comprehensively reviews various image-based techniques for skin disease detection, including traditional computer vision methods such as Convolutional Neural Network (CNN), Random Forest (RF), Naïve Bayes (NB), k-nearest Neighbour (k-NN), and Support Vector Machine (SVM). ISIC and dermofit datasets comprised 26,150 images divided into three categories: training (20,150), testing (3000), and validation (2000), respectively. Receiver Operating Characteristics (ROC) and Mean Squared Error (MSE) are used to calculate the accuracy and precision of models. As a result, it concluded that CNN outperformed with 94.91% accuracy. The proposed system can aid dermatologists in making accurate diagnoses and improve patient care.