Automated Classification of Skin Diseases Using CNNs and Outbreak Implications

Viruses are infectious organisms that are able to penetrate the human body and cause diseases, malfunctions, or dysfunctions of human organs. The present outbreak serves as a prime instance of how rapidly the virus spreads amongst people when they come into close, direct contact with lesions that contain it. Since the outbreak of May 2022, cases of monkeypox have been reported in diffierent countries where the illness is not prevalent. Among them, skin diseases affect millions of people worldwide, and accurate diagnosis plays a crucial role in effective treatment. Skin diseases can result from fungal infections, bacteria, allergies, viruses, and more. In this study, we gathered a dataset featuring images of three types of skin conditions: chickenpox, monkeypox, measles, and normal skin. The dataset was carefully curated, ensuring a diverse representation of skin conditions to capture the complexity and variations encountered in clinical practice. This paper presents, as its core, a system that exploits convolutional neural network (CNN) to classify colour images of skin lesions. The deep learning models rely on pre-trained weights from different CNN models such as VGG16, VGG19, InceptionV3, Xception, ResNet50, and DenseNet121. The data augmentation techniques are used to obtain the best possible outcomes. The categorical accuracy for VGG16, VGG19, InceptionV3, Xception, ResNet50, and DenseNet121 is 91.8%, 93.4%, 85.8%, 89.1%, 94.5% and 92.3%, respectively. Out of all the implemented models, the highest performance accuracy was obtained with DenseNet121, VGG19, and ResNet50.