NORMA eResearch @NCI Library

Using Chest X-ray Images to diagnose and distinguish COVID-19 Pneumonia, Viral Pneumonia, and Lung Opacity.

Inampudi, Gopi Krishna (2022) Using Chest X-ray Images to diagnose and distinguish COVID-19 Pneumonia, Viral Pneumonia, and Lung Opacity. Masters thesis, Dublin, National College of Ireland.

[thumbnail of Master of Science]
PDF (Master of Science)
Download (802kB) | Preview
[thumbnail of Configuration manual]
PDF (Configuration manual)
Download (327kB) | Preview


Diseases caused by coronaviruses are ubiquitous all over the world and have the potential to negatively impact not just human health but also the global economy. Coronavirus is a common respiratory virus that can cause pneumonia and can spread rapidly. The novel coronavirus COVID-19 is currently being fought around the globe. The globe was caught off guard by the disease's rapid spread. Even if a diagnostic kit for COVID-19 did exist, it would likely have a high false-negative rate (i.e., it would return a negative result even if the patient was infected with COVID-19), making it unavailable in most parts of the world. As a result, early diagnosis of COVID-19 is essential for reducing mortality and morbidity associated with the virus. Pneumonia symptoms are universal, and COVID-19 is no different. A chest X-ray is the gold standard for making a pneumonia diagnosis. As a result, detecting COVID-19 and the other anomalies it created has significantly boosted the demand for radiologists. In this paper, we propose a multi-class convolutional neural network model that uses transfer learning to automatically diagnose pneumonia and distinguish between COVID-19 and non-COVID-19 pneumonia. Input chest X-ray images into the model, and it can turn radiographic patterns into useful information and track structural changes in the lungs due to disease. Our model was developed using the COVID radiography dataset. We trained our model with 3 main alogorithms (Densenet121, InceptionV3 and Xception). The Xception model performed better than the other model in terms of Precision, recall and F1-score, with a classification accuracy of 88%. The simulation findings show that the suggested model is effective, can identify chest images rapidly and reliably, and aids doctors as a second reader in making a final choice.

Item Type: Thesis (Masters)
Subjects: Q Science > QA Mathematics > Electronic computers. Computer science
T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science
Q Science > QR Microbiology > QR355 Virology
H Social Sciences > HM Sociology > Information Science > Communication > Medical Informatics
Divisions: School of Computing > Master of Science in Data Analytics
Depositing User: Tamara Malone
Date Deposited: 26 Jan 2023 16:22
Last Modified: 26 Jan 2023 16:22

Actions (login required)

View Item View Item