Covid-19 Detection from Chest X-Ray Images and Hybrid Model Recommendation with Convolutional Neural Networks
The COVID-19 pandemic, which emerged at the end of 2019, continues to be effective. Although various vaccines have been developed, uncertainties remain over vaccine sharing, supply, storage and effect. The tendency of some countries to keep the developed vaccines only for their own citizens and usin...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Çanakkale Onsekiz Mart University
2021-12-01
|
| Series: | Journal of Advanced Research in Natural and Applied Sciences |
| Subjects: | |
| Online Access: | https://dergipark.org.tr/en/download/article-file/1825245 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The COVID-19 pandemic, which emerged at the end of 2019, continues to be effective. Although various vaccines have been developed, uncertainties remain over vaccine sharing, supply, storage and effect. The tendency of some countries to keep the developed vaccines only for their own citizens and using them as a political leverage shows that the pandemic will not end in the near future. In addition, discussions continue about the effectiveness of the proposed vaccine and drugs. For these reasons, the most effective method in the fight against COVID-19 is still considered to be using mask, social distance and 14-day isolation after disease detection. In most countries around the world, difficulties in diagnosing COVID-19 remain. Within the scope of the related study, the detection of COVID-19 from cost-effective and easily accessible lung X-Ray images was studied. The detection of COVID-19, which can be confused with other lung diseases from X-Ray images, can only be made by expert radiologists. In this context, a hybrid approach with high accuracy classification based on convolutional neural network has been proposed for the detection of COVID-19 pneumonia. In the proposed architecture, binary and multiple classification was made using MobileNetV2, DenseNet121, Inception ResNet V2 and Xception networks. Then, these networks were combined with stacking ensemble learning to create a hybrid model. |
|---|---|
| ISSN: | 2757-5195 |