COVID‐19 Detection Using Deep Convolutional Neural Networks and Binary Differential Algorithm‐Based Feature Selection from X‐Ray Images
Mohammad Saber Iraji, Mohammad-Reza Feizi-Derakhshi, Jafar Tanha
Abstract
The new COVID‐19 is rapidly spreading and has already claimed the lives of numerous people. The virus is highly destructive to the human lungs, and early detection is critical. As a result, this paper presents a hybrid approach based on deep convolutional neural networks that are very effective tools for image classification. The feature vectors were extracted from the images using a deep convolutional neural network, and the binary differential metaheuristic algorithm was used to select the most valuable features. The SVM classifier was then given these optimized features. For the study, a database containing images from three categories, including COVID‐19, pneumonia, and a healthy category, included 1092 X‐ray samples, was used. The proposed method achieved a 99.43% accuracy, a 99.16% sensitivity, and a 99.57% specificity. Our findings indicate that the proposed method outperformed recent studies on COVID‐19 detection using X‐ray images.