The author received no specific funding for this work.
Introduction: The early automatic diagnosis of the novel coronavirus (COVID-19) disease could be very helpful to reduce its spread around the world. In this study, we revisit the identification of COVID-19 from chest X-ray images using deep learning.
Methods: We collected a relatively large COVID-19 dataset—compared with previous studies—containing 309 real COVID-19 chest X-ray images. We also prepared 2,000 chest X-ray images of pneumonia cases and 1,000 images of healthy controls. Deep transfer learning was used to detect abnormalities in our image dataset. We fine-tuned three, pre-trained convolutional neural network (CNN) models on a training dataset: DenseNet 121, NASNetLarge, and NASNet-Mobile.
Results: The evaluation of our models on a test dataset showed that these models achieved an average sensitivity rate of approximately 99.45% and an average specificity rate of approximately 99.5%.
Conclusion: A larger dataset of COVID-19 X-ray images could lead to more accurate and reliable identification of COVID-19 infections using deep transfer learning. However, the clinical diagnosis of COVID-19 disease is always necessary.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Boudrioua, Mohamed Samir
"COVID-19 Detection from Chest X-ray Images using CNNs Models: Further Evidence from Deep Transfer Learning,"
The University of Louisville Journal of Respiratory Infections: Vol. 4
, Article 53.
Available at: https://ir.library.louisville.edu/jri/vol4/iss1/53