Yerkin Abdukarimov, Assanali Abu, M. Altynbekov, A. Shomanov, Seong-Jun Lee, Minho Lee
{"title":"Deep Neural Network Classification Models for COVID-19 Detection in X-ray Images","authors":"Yerkin Abdukarimov, Assanali Abu, M. Altynbekov, A. Shomanov, Seong-Jun Lee, Minho Lee","doi":"10.1109/icecco53203.2021.9663823","DOIUrl":null,"url":null,"abstract":"At the beginning of 2020 new COVID-19 infection became a global pandemic, and society needed an efficient method to detect infected people. To handle the spread of infection testing systems were developed. But due to the fact that they take a lot of time and are not available to everyone, alternative methods of early screening have become an urgent need. In our paper, we propose to use convolutional neural networks (CNN) to detect coronavirus infection on X-ray images. We have collected 9 of the most popular datasets containing x-ray images of patients infected with COVID-19 or pneumonia and classified on most common CNN models: ResNet50, VGG- 16, Alexnet, Inception-v3, and InceptionResNet-v2. Based on results we obtained it was possible to generate a heat map that indicates areas containing features that distinguish infected patients most effectively. Also, 2D T-SNE images were created to provide a lower dimensional overview of the data distribution among 2 classes representing infected scans vs normal scans. In our experiments, the InceptionResNet-v2 model showed best test result and the average prediction value reached 95.1%, which is a very promising accuracy for classifying healthy and infected patients.","PeriodicalId":331369,"journal":{"name":"2021 16th International Conference on Electronics Computer and Computation (ICECCO)","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2021-11-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"5","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2021 16th International Conference on Electronics Computer and Computation (ICECCO)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/icecco53203.2021.9663823","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 5
Abstract
At the beginning of 2020 new COVID-19 infection became a global pandemic, and society needed an efficient method to detect infected people. To handle the spread of infection testing systems were developed. But due to the fact that they take a lot of time and are not available to everyone, alternative methods of early screening have become an urgent need. In our paper, we propose to use convolutional neural networks (CNN) to detect coronavirus infection on X-ray images. We have collected 9 of the most popular datasets containing x-ray images of patients infected with COVID-19 or pneumonia and classified on most common CNN models: ResNet50, VGG- 16, Alexnet, Inception-v3, and InceptionResNet-v2. Based on results we obtained it was possible to generate a heat map that indicates areas containing features that distinguish infected patients most effectively. Also, 2D T-SNE images were created to provide a lower dimensional overview of the data distribution among 2 classes representing infected scans vs normal scans. In our experiments, the InceptionResNet-v2 model showed best test result and the average prediction value reached 95.1%, which is a very promising accuracy for classifying healthy and infected patients.