Show simple item record

AuthorMrad, Ilyes
AuthorHamila, Ridha
AuthorErbad, Aiman
AuthorHamid, Tahir
AuthorMazhar, Rashid
AuthorAl-Emadi, Nasser
Available date2023-04-04T09:09:09Z
Publication Date2021
Publication NameProceedings - European Workshop on Visual Information Processing, EUVIP
ResourceScopus
URIhttp://dx.doi.org/10.1109/EUVIP50544.2021.9484001
URIhttp://hdl.handle.net/10576/41644
AbstractCOVID-19 is a virus that has infected more than one hundred and fifty million people and caused more than three million deaths by 13th of Mai 2021 and is having a catastrophic effect on the world population's safety. Therefore, early detection of infected people is essential to fight this pandemic and one of the main screening methods is radiological testing. The goal of this study is the usage of chest x-ray images (CXRs) to effectively identify patients with COVID-19 pneumonia. To achieve an efficient model, we combined three methods named: Convolution Neural Network (CNN), transfer learning, and the focal loss function which is used for imbalanced classes to build 3 binary classifiers, namely COVID-19 vs Normal, COVID-19 vs pneumonia and COVID-19 vs Normal Pneumonia (Normal and Pneumonia). A comparative study has been made between our proposed classifiers with well-known classifiers and provided enhanced results in terms of accuracy, specificity, sensitivity and precision. The high performance of this computer-Aided diagnostic technique may greatly increase the screening speed and reliability of COVID-19 detection. 2021 IEEE.
SponsorThis work was supported by Qatar University Internal Grant IRCC-2020-001. The statements made herein are solely the responsibility of the authors.
Languageen
PublisherIEEE
Subjectchest X-ray images
convolutional neural network
COVID-19
focal loss function
TitleMachine Learning Screening of COVID-19 Patients Based on X-ray Images for Imbalanced Classes
TypeConference Paper
Volume Number2021-June
dc.accessType Open Access


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record