Enhancing COVID-19 Detection: An Xception-Based Model with Advanced Transfer Learning from X-ray Thorax Images

Area of Science:

• Medical Imaging
• Artificial Intelligence in Healthcare
• Infectious Disease Diagnostics

Background:

• Accurate and rapid Coronavirus Disease 2019 (COVID-19) detection is critical for patient management and pandemic control.
• Existing deep learning models, particularly Convolutional Neural Networks (CNNs), face challenges like overfitting and high computational costs in medical image analysis.
• There is a need for more efficient and accurate AI models for COVID-19 diagnosis using radiological images.

Purpose of the Study:

• To develop and evaluate an innovative deep learning model for enhanced COVID-19 detection from chest X-ray images.
• To address the limitations of existing models, including expressiveness issues and resource-intensive training.
• To improve the accuracy and efficiency of AI-driven COVID-19 diagnostic tools.

Main Methods:

• Utilized the Xception architecture, a state-of-the-art CNN, for image classification.
• Incorporated advanced transfer learning techniques to augment the Xception model's performance.
• Trained and validated the model on a dataset of chest X-ray images for COVID-19 identification.

Main Results:

• The proposed Xception model with transfer learning demonstrated superior predictive accuracy compared to baseline models like VGG-16 and ResNet.
• The model effectively identified COVID-19 cases from chest X-ray images, outperforming established methods.
• Experimental results indicate enhanced diagnostic performance and potential for overcoming common deep learning challenges.

Conclusions:

• The developed transfer learning-enhanced Xception model offers a significant advancement in AI-based COVID-19 detection from X-rays.
• This approach represents a promising stride towards more accurate, efficient, and accessible diagnostic tools for infectious diseases.
• The findings suggest a viable solution for improving patient outcomes and public health surveillance during pandemics.