A Deep Learning Approach for COVID-19 8 Viral Pneumonia Screening with X-ray Images

Abstract

Beginning in December 2019, the spread of the novel Coronavirus (COVID-19) has exposed weaknesses in healthcare systems across the world. To sufficiently contain the virus, countries have had to carry out a set of extraordinary measures, including exhaustive testing and screening for positive cases of the disease. It is crucial to detect and isolate those who are infected as soon as possible to keep the virus contained. However, in countries and areas where there are limited COVID-19 testing kits, there is an urgent need for alternative diagnostic measures. The standard screening method currently used for detecting COVID-19 cases is RT-PCR testing, which is a very time-consuming, laborious, and complicated manual process. Given that nearly all hospitals have X-ray imaging machines, it is possible to use X-rays to screen for COVID-19 without the dedicated test kits and separate those who are infected and those who are not. In this study, we applied deep convolutional neural networks on chest X-rays to determine this phenomena. The proposed deep learning model produced an average classification accuracy of 90.64% and F1-Score of 89.8% after performing 5-fold cross-validation on a multi-class dataset consisting of COVID-19, Viral Pneumonia, and normal X-ray images.

Chest X-ray analysis empowered with deep learning: A systematic review

“…Once the model has been selected and trained using the prepared dataset, the model can overfit. To overcome this, techniques like regularization, early stopping and feature removing can be tried out [23] , [42] , [75] , [113] . Finally, model refinements can be applied to improve the performance of the trained model.…”

“…For instance, Ahmed et al. [42] have presented a CNN model with five convolutional layers, each followed by batch normalization and max-pooling layers, and a dropout where the final layer is fully connected. They have used RMSProp as the optimizer and showed an overall accuracy of 90.64% in detecting COVID-19 and Pneumonia.…”

Application of Machine Learning and Deep Learning Techniques for COVID-19 Screening Using Radiological Imaging: A Comprehensive Review

“…Several researchers have been experimenting with various ResNet architectural variants as a result of their findings [ 244 – 246 ]. Along with pre-train deep learning architectures, some other articles used only CNN models with different types of layer sizes [ 247 – 249 ]. To improve the convolution neural network performance for identifying COVID-19, geometric image augmentation is one of the essential factors [ 250 , 251 ].…”

Artificial intelligence approaches and mechanisms for big data analytics: a systematic study

“…Also, a deep CNN on chest X-rays is proposed by Ahmed, Bukhari & Keshtkar (2021) to determine COVID-19. After 5-fold cross-validation on a multi-class dataset consisting of COVID-19, Viral Pneumonia, and normal X-ray images, the proposed method achieved a classification accuracy of 90.64%.…”