Ferdi, AbdesselamBenierbah, SaidFerdi, Youcef2025-04-172025-04-172021http://dspace.univ-oeb.dz:4000/handle/123456789/21900In recent years, deep learning algorithms have acquired a great popularity in many fields including medical imaging to achieve different tasks such as image segmenta-tion, classification, detection, and retrieval tasks. Although reverse transcription po-lymerase chain reaction test (RT-PCR) is considered to be the gold standard for CO-VID- 19 screening, deep learning algorithms may help radiologists to automatically diagnose this disease by processing chest Xray (CXR) and chest CT images. This paper proposes a comparative study of five deep learning models for classification of CXR COVID-19 images. The five selected models are pre-trained networks, namely, ResNet-50, InceptionResNet-v2, Inception-v3, MobileNet-v2, and GoogleNet. The models were evaluated using CXR COVID-19 and normal images from publicly avail-able datasets. The obtained classification results show that the GoogleNet model achieves the highest classification performance with a validation accuracy of 95.69%.enCOVID-19 ; Deep Learning ; Convolutional Network ; Chest X-ray image ; Medical Image ProcessingArticle