Covid-19 and other types of pneumonia are diagnosed with high accuracy from chest X-ray images with the support of Artificial Intelligence

Yayın Tarihi | 23 May 2024, Thursday

Associate Professor Dr. Onur SEVLI, a faculty member at Burdur Mehmet Akif Ersoy University, has managed to achieve high accuracy results for the diagnosis of COVID-19 and other types of pneumonia from chest X-rays (CXR) using deep learning-based models. In the study, a model developed specifically for the study was used alongside modern Convolutional Neural Network (CNN) architectures to distinguish COVID-19 from other types of pneumonia.


In the research, a data set consisting of a total of 15,153 CXR images was used. This data set consists of different classes, including COVID-19, other pneumonias, and healthy lung images. As a result of the evaluation of different CNN models, the ResNet18 architecture showed higher success compared to other models. The findings obtained in the study provided an important reference point when compared with existing studies in the literature.


A pilot test study was conducted with 10 experts to test the practical benefit of the model. In this test, the accuracy and usability of the model were evaluated. The test results confirmed the potential of the model as a reliable decision support system in the diagnosis of COVID-19. Feedback from experts reinforced the applicability and accuracy of the model in practical diagnostic processes.


In future studies, it is aimed to improve the parameters of the model and test it with larger data sets. In addition, it is planned to improve diagnostic results by diversifying data set samples and conducting broader application tests. These studies play a significant role in reducing the effects of the disease by contributing to the early diagnosis and timely treatment of COVID-19.


The research emphasizes the importance of artificial intelligence-based decision support systems, especially to meet the demand for effective health solutions during the pandemic. This innovative CNN approach aims to optimize the accuracy of COVID-19 diagnosis and improve treatment processes. The full text of the study has been published in the relevant scientific journal, and more detailed information can be accessed.


Link to the Study: