Artificial Intelligence-Based Classification of Chest X-Ray Images into COVID-19 and Other Infectious Diseases

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Infectious Diseases

Background:

• The COVID-19 pandemic presents significant healthcare challenges, necessitating rapid patient identification and monitoring.
• Current diagnostic methods like RT-PCR can be time-consuming, driving research into faster alternatives.
• Efficient screening is crucial for timely treatment and management of COVID-19 patients.

Purpose of the Study:

• To create efficient deep learning models for rapid COVID-19 screening using chest X-ray images.
• To develop Artificial Intelligence (AI)-based classification models for COVID-19 and other major infectious diseases.
• To evaluate the efficacy of AI models in classifying various conditions from chest X-rays.

Main Methods:

• Utilized publicly available PA chest X-ray images of adult COVID-19 patients.
• Applied 25 different data augmentation techniques to increase dataset size and model generalizability.
• Employed a transfer learning approach for training and testing AI classification models.
• Combined two best-performing models trained on augmented image datasets.

Main Results:

• The developed AI models demonstrated high prediction accuracy in classifying normal, COVID-19, non-COVID-19, pneumonia, and tuberculosis from chest X-rays.
• The combination of optimized models achieved superior performance compared to previously published methods.
• Transfer learning significantly enhanced the efficiency and accuracy of AI-based image classification.

Conclusions:

• AI-based classification models utilizing transfer learning can efficiently classify chest X-ray images for various diseases, including COVID-19.
• This approach offers a promising, efficient alternative or supplement to conventional diagnostic methods.
• The study represents a significant step towards implementing AI in biomedical imaging for COVID-19 and related conditions.