E ISSN: 2583-049X

International Journal of Advanced Multidisciplinary Research and Studies

Volume 3, Issue 5, 2023

Taxonomy of COVID-19 Disease Radiology by using Convolutional Neural Networks

Author(s): Arkan Abass Mohammed, Abdolvahab Ehsani Rad


The viral illness known as coronavirus disease (COVID-19) is mostly caused by the SARS-CoV-2 virus. The majority of individuals who get the virus will have a mild to severe respiratory infection and achieve recovery without necessitating specialized medical intervention. Nevertheless, a portion of individuals will have severe illness and require medical intervention. Individuals who are advanced in age or possess pre-existing medical diseases such as cardiovascular disease, diabetes, chronic respiratory disease, or cancer have a higher propensity for the development of severe sickness. Individuals of all ages are susceptible to contracting COVID-19 and experiencing severe illness or mortality.

This research utilizes global datasets and employs deep learning techniques, namely Convolutional Neural Networks (CNNs), to enhance the obtained results. The model's validation is conducted using a Global dataset named "COVID-19Radiography" sourced from Kaggle. The performance of the model may be summarized as follows: during the training phase, the accuracy achieved was 100.0% and the loss metric was measured to be 0.0349. In the validation phase, the accuracy reached 99% and the loss metric was recorded as 0.0340. As a consequence, a model was developed and transformed into a graphical user interface program named "AzadCXR Covid19" with a focus on user-friendliness. Clinicians and radiologists possess the proficiency to effectively use and utilize chest X-ray (CXR) images for the purpose of diagnosing individuals with suspected cases of COVID-19.

Keywords: Deep Learning Techniques, CNNs, COVID-19 Radiography, Chest X-Ray (CXR)

Pages: 655-662

Download Full Article: Click Here