Publication: Pneumonia and Covid-19 Detection on Chest X-ray Imaging Using Deep Learning: PCX-Net
2
0
Issued Date
2025-01-01
Resource Type
Scopus ID
2-s2.0-105007163186
Journal Title
Proceedings Ieecon 2025 2025 13th International Electrical Engineering Congress Carbon Neutrality Challenges and Solutions Based on Sustainable Power of Nature
Rights Holder(s)
SCOPUS
Bibliographic Citation
Proceedings Ieecon 2025 2025 13th International Electrical Engineering Congress Carbon Neutrality Challenges and Solutions Based on Sustainable Power of Nature (2025)
Suggested Citation
Sa-Ing V., Srisomboon K., Lee W., Kaewtreewong W., Thongruy K., Chaysangchom M. Pneumonia and Covid-19 Detection on Chest X-ray Imaging Using Deep Learning: PCX-Net. Proceedings Ieecon 2025 2025 13th International Electrical Engineering Congress Carbon Neutrality Challenges and Solutions Based on Sustainable Power of Nature (2025). doi:10.1109/iEECON64081.2025.10987843 Retrieved from: https://hdl.handle.net/20.500.14740/21083
Author's Affiliation
Corresponding Author(s)
Other Contributor(s)
Abstract
Pneumonia and COVID-19 are severe diseases that primarily affect the lungs, causing acute and significant respiratory distress. In clinical settings, patients are screened to differentiate between these two conditions using computerized chest X-ray imaging. To address this issue, researchers have explored the solutions based on deep learning model to automatic analysis on the chest X-ray images. This research evaluates the four neural network models consist of VGG16, ResNet50, MobileNet, and Xception to classify the chest X-ray images on three categories: pneumonia, COVID-19, and normal lungs. This research involved preprocessing three datasets of chest X-ray images, including artifact removal, image resizing, and data augmentation, prior to model training and testing processing. From our experimental results, Xception model demonstrated the highest performance, achieving training accuracy of 99.99%, validation accuracy of 98.00%, training loss of 0.15%, and validation loss of 7.42%, with an overall accuracy of 97.70%. Thus, the Xception model is the most effective for distinguishing and predicting lung conditions from chest X-ray images and should be considered for future applications in medical data analysis.