Deep Learning Models for the Detection and Classification of COVID-19 and Associated Lung Diseases Using X-Ray Images
| dc.contributor.author | Dikmen, Osman | |
| dc.date.accessioned | 2025-03-24T19:50:00Z | |
| dc.date.available | 2025-03-24T19:50:00Z | |
| dc.date.issued | 2024 | |
| dc.department | Düzce Üniversitesi | |
| dc.description.abstract | The COVID-19 pandemic has introduced exceptional challenges to healthcare systems worldwide, underscoring the urgent need for swift and precise diagnostic solutions. In this research, we investigate the performance of various deep learning models, including VGG19, ResNet18, and a ResNet18-based U-Net, as well as a Custom Convolutional Neural Network (CNN) developed in MATLAB, for the classification and segmentation of lung X-ray images. The dataset includes X-ray images from individuals diagnosed with COVID-19, viral pneumonia, lung opacity, and healthy individuals. The dataset was divided into 80% for training and 20% for testing, with data augmentation techniques implemented to enhance the model's effectiveness. The VGG19 model, utilizing transfer learning, demonstrated strong diagnostic capabilities, achieving high accuracy rates for COVID-19, lung opacity, healthy lungs, and viral pneumonia classification, with a test accuracy of 97.5%. ResNet18 was employed for both classification and as part of a hybrid model incorporating a U-Net-inspired decoder for lung disease segmentation. The ResNet18 model achieved competitive accuracy and loss metrics, while the ResNet18-based U-Net model excelled in image segmentation tasks, demonstrating its potential in biomedical image analysis. Additionally, a Customized CNN model was developed using MATLAB for the classification of the four lung conditions. This model showed visual outputs including training-validation loss/accuracy graphs and confusion matrices. Our results indicate that deep learning models, especially when combined with transfer learning and customized architectures, offer a powerful approach to diagnosing COVID-19 and related lung conditions. Future work will focus on refining these models with larger datasets and further experimentation to enhance diagnostic performance across diverse clinical settings. | |
| dc.identifier.endpage | 142 | |
| dc.identifier.issn | 2757-9778 | |
| dc.identifier.issue | 2 | |
| dc.identifier.startpage | 121 | |
| dc.identifier.uri | https://hdl.handle.net/20.500.12684/19407 | |
| dc.identifier.volume | 4 | |
| dc.language.iso | en | |
| dc.publisher | İzmir Bakırçay Üniversitesi | |
| dc.relation.ispartof | Artificial Intelligence Theory and Applications | |
| dc.relation.publicationcategory | Makale - Ulusal Hakemli Dergi - Kurum Öğretim Elemanı | |
| dc.rights | info:eu-repo/semantics/openAccess | |
| dc.snmz | KA_DergiPark_20250324 | |
| dc.subject | COVID-19|Deep Learning|Medical Image Classification|X-Ray Imaging|Convolutional Neural Networks (CNN). | |
| dc.title | Deep Learning Models for the Detection and Classification of COVID-19 and Associated Lung Diseases Using X-Ray Images | |
| dc.type | Article |
