Cost-Effective Image based Skin Disease Detection in Resource-Constrained Settings using Machine Learning

In regions like India, where skin diseases are prevalent due to hot weather and desert conditions, this research study presents a cost-effective solution for early detection and classification of skin diseases using image processing techniques. Leveraging digital images of affected skin areas, the proposed approach utilizes image analysis, including feature extraction through a pretrained Regional Convolutional Neural Network (RCNN), followed by Support Vector Machine (SVM) classification. The system achieves a commendable 82% accuracy rate in identifying Melanoma, a particularly dangerous form of skin disease. With the scarcity of medical knowledge among the general public and the potential delay in symptom recognition, this simple and efficient method, requiring only a camera and a computer, can bridge the diagnostic gap, aiding in timely disease identification and severity assessment. The dataset compilation adheres to ethical considerations, ensuring patient privacy. By addressing issues related to database creation and image dimensions, the research contributes to enhancing the performance of skin disease detection systems, ultimately offering an accessible and impactful tool for healthcare professionals and individuals alike.