| dc.creator |
Kaya, Furkan |
|
| dc.creator |
TÜRKÇETİN, Ayşen Özün |
|
| dc.creator |
KOÇ, Turgay |
|
| dc.creator |
Çilekar, Şule |
|
| dc.creator |
Inkaya, Yaşar |
|
| dc.creator |
Konya, Petek Şarlak |
|
| dc.date |
2023-01-01T00:00:00Z |
|
| dc.date.accessioned |
2025-02-25T10:38:04Z |
|
| dc.date.available |
2025-02-25T10:38:04Z |
|
| dc.identifier |
c6c39bf8-f521-4432-b4d6-6956fe3c6f13 |
|
| dc.identifier |
10.1109/eleco60389.2023.10416075 |
|
| dc.identifier |
https://avesis.sdu.edu.tr/publication/details/c6c39bf8-f521-4432-b4d6-6956fe3c6f13/oai |
|
| dc.identifier.uri |
http://acikerisim.sdu.edu.tr/xmlui/handle/123456789/101311 |
|
| dc.description |
Although it is known that the COVID-19 process is over, the subsequent damage caused by COVID-19 on the body is undeniable. The SARS-CoV-2 virus has been shown to be responsible for causing acute respiratory distress in a large number of COVID-19 cases. When the literature is reviewed, Thorax Computed Tomography is recommended to evaluate permanent lung damage in individuals recovering from COVID-19. When tomography images are examined, the respiratory systems of patients who have had COVID-19 are significantly affected by the virus. In the study, thorax CT images of patients who had COVID-19 and then came back to the hospital and were diagnosed with COPD and Bronchiectasis were examined with artificial intelligence methods. The study consists of two stages. First, COVID-19, COPD, and Bronchiectasis datasets were trained with pre-trained deep-learning models. Then, only the COVID-19 dataset was trained with the GAN algorithm, and the model weights were recorded. Tests were performed on the COPD and Bronchiectasis dataset with the recorded model weight. The performance metric ratios of the training results obtained with the COVID-19 dataset and test trained individually and the combination of three classes was compared in the test environment. As a result of the proposed study, the highest performance rate for the classification method made in the first step was found to be 99% in the pre-trained DenseNet201 model. In the second step of the study, model weights were recorded on the COVID-19 dataset trained with the GAN algorithm and tested on COPD and Bronchiectasis datasets. Lower loss rates are observed in the GAN algorithm compared to the models in the first stage of the study. |
|
| dc.language |
eng |
|
| dc.rights |
info:eu-repo/semantics/closedAccess |
|
| dc.title |
Artificial Intelligence Approach in the Detection of Lung Diseases Developing Post-COVID-19 with Lung Images |
|
| dc.type |
info:eu-repo/semantics/conferenceObject |
|