Prognosticating COVID-19 Pneumonia Versus Common Pneumonia for Pediatrics Using Chest Radiographic Images

There is currently no authorized medication or vaccination for the recently identified coronavirus, COVID-19, which is extremely contagious and causes illnesses. Omicron variant is the most transmissible variant for all age groups especially for children. The most typical COVID-19 symptoms include a coughing, fever, and sore throat. When symptoms intensify into a severe form of pneumonia, acute breathing crisis septic shock, pulmonary edema, and multiple organ failure may happen. Identifying the different pneumonias is difficult for children. To identify the most precise, crucial element of learning, deep convolutional neural networks, DenseNet, ResNet, MobileNet, Xception, InceptionResNetV2, VGGNet, InceptionV3, and NASNet were selected from deep CNN. The collected features are then used to classify participants as COVID-19 modules or controls using different machine learning separators. To develop a robust capacity for the normalization of anonymous data, this strategy has avoided methods of preprocessing specific work data. The effectiveness of the proposed method is evaluated based on publicly available Chest X-ray (CXR) images of COVID-19. The deep CNN is used to extract the features, which are integrated with the addition of Generative Adversarial Network (CYCLE-GAN). The binary classification will be made into a hybrid Deep CNN-LSTM proposed model. This model classifies the images into COVID-19 pneumonia, common pneumonia, and no pneumonia. It achieves an accuracy of 96.3% and precision of 94.5% and F1-score of 91.3%. This effective classification helps the medical practitioner to detect the type pneumonia and its severity clearly and further go for the appropriate medical treatment.