DEVELOPMENT OF AN INTELLIGENT DIAGNOSTIC SYSTEM FOR THE DETECTION OFSKIN DISEASES USING ARTIFICIAL INTELLIGENCE TECHNIQUE

Abstract

The worldwide occurrence of skin ailments emphasises the necessity for effective and precise diagnostic tools. To categorize skin disorders from clinical images, this study presents a unique automated skin disease detection system that integrates image processing approaches with machine learning, particularly a Convolutional Neural Network (CNN). Preprocessing techniques, including noise reduction, colour correction, and segmentation, are all part of the suggested approach, which aims to improve picture quality and guarantee precise feature extraction. The system collects important aspects necessary for accurate illness classification, such as color, texture, and shape, using a 2D Wavelet Transform algorithm and a Structural Co-Occurrence Matrix (SCM). 1,880 skin photographs from various patient populations, including both healthy and diseased skin conditions, were used to train and verify the system. With a multi-class classification accuracy of 87.9% and a binary classification accuracy of 98.3%, the CNN model demonstrated outstanding classification accuracy. The model's performance was further enhanced by adding patient data, and it was able to achieve an overall accuracy of 97.5% on unseen data