COVID-19 Classification on Chest X-ray Images Using Deep Learning Methods.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Pulmonology

Background:

• The coronavirus disease (COVID-19) pandemic has caused a global health crisis.
• Chest radiography (CXR) is crucial for monitoring COVID-19's pulmonary effects and reducing mortality.
• Deep learning (DL) models have shown potential for COVID-19 diagnosis from CXR.

Purpose of the Study:

• To evaluate the performance of five deep learning models for COVID-19 detection using chest X-ray images.
• To highlight the potential of individual DL models in analyzing COVID-19 CXR.
• To assess the diagnostic capabilities of DL models in a real-world medical context.

Main Methods:

• Utilized ResNet50, ResNet101, DenseNet121, DenseNet169, and InceptionV3 models with Transfer Learning.
• Trained and validated models on a large, publicly available COVID-19 CXR dataset.
• Evaluated model performance on unseen data to ensure generalizability.

Main Results:

• All analyzed deep learning models demonstrated satisfactory performance.
• ResNet101 achieved superior results with 96% Precision, Recall, and Accuracy.
• The models showed significant potential for COVID-19 diagnosis from CXR.

Conclusions:

• Deep learning models offer a promising approach for understanding and diagnosing COVID-19.
• ResNet101 stands out as a highly effective model for COVID-19 detection in CXR.
• These findings support the integration of AI in medical imaging for infectious disease management.