Monkeypox Detection Using Particle Swarm Optimization‑Based Extreme Learning Machine

Abstract

Monkeypox is caused by a virus that causes skin lesions and has an average mortality rate of 11% in unvaccinated patients. This study proposes a particle swarm optimization-based extreme learning machine (PSO-ELM) model to detect monkeypox disease. First, monkeypox images were read from two public databases. Secondly, image preprocessing is implemented, including noise removal, contrast enhancement, resizing, rotation and flipping. Third, a convolution neural network (CNN) model is then used to extract the required features from the processed images. Finally, the features extracted by the CNN were fed into PSO-ELM to distinguish monkeypox, chickenpox, and measles. A ten-fold cross-validation method is used to evaluate the performance of the proposed system. Experimental results show that the accuracy, precision, recall and F1-score of PSO-ELM are 0.961, 0.969, 0.961 and 0.964 respectively, which is better than the classic CNN. Moreover, the average training time of PSO-ELM is only about 16 seconds, which is significantly lower than classic CNN. This research result can be applied in clinical detection of monkeypox.