Detection of coronavirus disease from X-ray images using deep learning and transfer learning algorithms

Area of Science:

• Artificial Intelligence
• Medical Imaging
• Computer Vision

Background:

• The COVID-19 pandemic highlighted the need for rapid and accurate diagnostic tools.
• Medical image analysis, particularly with chest X-rays, shows promise for identifying pneumonia and related conditions.

Purpose of the Study:

• To develop an automated model for early detection of COVID-19 using computer vision and medical image analysis.
• To leverage transfer learning and deep learning techniques for COVID-19 diagnosis from chest X-ray images.

Main Methods:

• Applied transfer learning to fine-tune three deep learning models: Inception ResNetV2, InceptionNetV3, and NASNetLarge.
• Trained models on a dataset of 850 COVID-19 positive, 500 pneumonia, and 915 normal chest X-ray images.
• Evaluated model performance based on accuracy, with and without data augmentation.

Main Results:

• InceptionNetV3 demonstrated superior performance, achieving 98.63% accuracy with data augmentation and 99.02% without.
• All models exhibited a tendency towards overfitting when data augmentation was not utilized, attributed to limited training data.
• The study confirmed the feasibility of deep transfer learning for automated COVID-19 detection.

Conclusions:

• Deep transfer learning models can effectively detect COVID-19 from chest X-rays, assisting clinicians in decision-making.
• The study provides insights into applying transfer learning for automated disease detection.
• Future research should explore advanced convolutional neural network models with larger datasets for enhanced efficiency.