A Deep Learning Model for Diagnosing COVID-19 and Pneumonia through X-ray

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Infectious Diseases

Background:

• The COVID-19 pandemic necessitates rapid diagnostic tools.
• Existing detection methods like RT-PCR have limitations in speed and accessibility.
• Timely diagnosis is crucial for controlling the spread of COVID-19.

Purpose of the Study:

• To develop a rapid and practical method for COVID-19 detection.
• To explore the application of deep learning in diagnosing COVID-19 from chest X-ray (CXR) images.
• To create an ensemble deep learning model for accurate classification of pneumonia and COVID-19.

Main Methods:

• Feature selection from chest radiographs.
• Development of a CXR classification method.
• Application of transfer learning and ensemble learning.
• Proposed ensemble deep learning model: COVID-ensemble.

Main Results:

• The proposed feature selection method improved CNN model accuracy.
• COVID-ensemble achieved 91.5% accuracy in binary classification and 91.11% in multi-category classification.
• The model demonstrated superior performance compared to existing methods.

Conclusions:

• COVID-ensemble provides a fast and convenient alternative to RT-PCR for COVID-19 detection.
• The model can automatically detect pneumonia, alleviating medical staff workload.
• Deep learning on X-ray images offers an efficient screening method for COVID-19 and pneumonia.