2023 •
A binary classification model of COVID-19 based on convolution neural network
Authors:
Reham Sabah Saeed, Bushra Kadhim Oleiwi Chabor Alwawi
Abstract:The outbreak of the new coronavirus (COVID-19) had resulted in the creation of a disaster all over the world and it had become a highly acute and severe illness. The prevalence of this disease is increasing rapidly worldwide. The technology of deep learning (DL) became one of the hot topics in the computing context and it is widely implemented in a variety of the medical applications. Those techniques proved to be sufficient tools for the clinicians in automatic COVID-19 diagnosis. In the present study, a DL technology that is based on (...) The outbreak of the new coronavirus (COVID-19) had resulted in the creation of a disaster all over the world and it had become a highly acute and severe illness. The prevalence of this disease is increasing rapidly worldwide. The technology of deep learning (DL) became one of the hot topics in the computing context and it is widely implemented in a variety of the medical applications. Those techniques proved to be sufficient tools for the clinicians in automatic COVID-19 diagnosis. In the present study, a DL technology that is based on convolution neural networks (CNN) models had been suggested for the binary COVID-19 classification. In the initial step of the suggested model, COVID-19 data-set of chest X-ray (CXR) images have been obtained then preprocessed. Whereas in the second stage, a new CNN model has been built and trained for diagnosing COVID-19 data-set as (positive) infection or (negative) normal cases. The suggested architecture had a success in classifying COVID-19 with the training model accuracy that had reached 96.57% for the training data-set and 92.29% for validating data-set and could reach the target point with a minimal learning rate for training this model with promising results.(Read More)
We have placed cookies on your device to help make this website and the services we offer better. By using this site, you agree to the use of cookies. Learn more