Comparison Results of Hybrid CNN-Machine Learning Algorithms Architectures for Monkeypox Images Classification

Monkeypox is a rare infectious viral disease that occurs due to a particular virus that may be transferred from infected animals to other humans or even between people themselves. This disease is characterized by several symptoms similar to those observed in patients with original smallpox, but less severe. In 2022 the World Health Organization (WHO) proclaimed this illness as a worldwide emergency public health issue, and this is as a consequence of its propagation in several countries of the world beyond its borders in western and central Africa. In most cases, the monkeypox virus heals spontaneously and has a very low overall case fatality rate. However, its diagnosis may be confused with other rash diseases and its confirmation may require biological tests. This paper is a comparative study between several hybrid architectures composed of both deep learning models and machine learning methods that have been implemented on a dataset constituted of a collection of 228 images, including multiple positive and negative cases of the monkeypox virus. The majority of the developed models have recorded significant performance metrics, reaching for some models an accuracy and an F1-score of 94% which may facilitate the early detection of suspicious cases and reduce the extensive propagation of the disease.