Publication Date: 2024/02/16
Abstract: The onset of the Coronavirus Disease 2019 (COVID-19) outbreak in early December 2019 has had profound and far-reaching repercussions on global public health. Despite being the gold standard for diagnosis, reverse transcription-polymerase chain reaction (RT- PCR) alone is unable to address the pandemic’s urgent need for rapid and efficient diagnostic methods because of its time-consuming and complex nature. In this study, we propose a novel convolutional neural network (CNN) model, which is trained with a publicly available dataset, with targets of the normal, COVID-19, and viral pneumonia classes. The trained model achieved accuracy of 97.17% and specific recall of 94% in COVID-19 cases. A web application developed using the Python Flask framework is developed, whereby the users are able to upload X-ray images and acquire the prediction results and gradient activation map (Grad-CAM) of the images. This web app can help to provide a second opinion to medical practitioners regarding COVID-19 diagnosis.
Keywords: CNN, COVID-19 Diagnosis, GradCAM, Web Application, X-ray İmages.
DOI: https://doi.org/10.5281/zenodo.10669604
PDF: https://ijirst.demo4.arinfotech.co/assets/upload/files/IJISRT24FEB411.pdf
REFERENCES