Résumé
The novel Corona Virus (COVID-19) has spread so rapidly that cause a devastating effect on public well-being and create an emergency around the world. Hence, the rapid identification of COVID-19 has become a challenging work within a short period. Clinical trials of patients with COVID-19 have shown that most of the patients affected by COVID-19 experience lung infection that can cause inflammation in the lung after virus-contiguity. It can damage the cells and tissue that is inside the lung. However, pneumonia is also a lung infection that can cause inflammation in the air sacs inside the lung. Chest X-rays and CT scans perform an essential role in the detection of lung-related illnesses. Therefore, concerning the diagnosis of COVID-19, radiography and chest CT are considered as fundamental imaging approaches. This study presents a densely interconnected convolutional neural network-based approach to identify COVID-19, Pneumonia and Normal patients from chest X-ray images. To experiment with the proposed methodology, a new dataset is generated by combining two different datasets from Kaggle named COVID-19 Radiography Database and Chest X-ray (COVID-19 & Pneumonia). The dataset comprises of 500 X-ray images of COVID-19 affected people, 2600 X-ray images of Normal people, and 3418 X-ray images of pneumonia affected people. The proposed densely interconnected convolutional neural network model produces 99% testing accuracy for COVID-19, 98% testing accuracy for Pneumonia and 98% testing accuracy for Normal people without the application of any augmentation techniques. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.