Randomly initialized convolutional neural network for the recognition of COVID-19 using X-ray images

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Infectious Diseases

Background:

• The COVID-19 pandemic necessitates rapid and accurate diagnostic tools.
• Deep Learning (DL) models, particularly Convolutional Neural Networks (CNNs), show promise for analyzing medical images like chest X-rays for disease detection.
• Existing CNN models often have limitations in COVID-19 detection accuracy.

Purpose of the Study:

• To propose and evaluate a novel randomly initialized CNN (RND-CNN) architecture for the recognition of COVID-19 from chest X-ray images.
• To assess the performance of the RND-CNN model on public datasets for COVID-19 detection.

Main Methods:

• Development of a novel RND-CNN architecture with custom-sized hidden layers.
• Evaluation of the RND-CNN model using two public datasets: COVIDx and an enhanced COVID-19 dataset.
• Classification of chest X-ray images into three categories: COVID-19 positive, pneumonia, or normal.

Main Results:

• The RND-CNN model demonstrated high accuracy in identifying COVID-19 from chest X-rays.
• Achieved 94% accuracy on the COVIDx dataset.
• Achieved 99% accuracy on the enhanced COVID-19 dataset.

Conclusions:

• The proposed RND-CNN architecture is effective for COVID-19 detection using chest X-ray analysis.
• The model's high accuracy suggests its potential as a valuable tool in combating the COVID-19 pandemic.
• Further research can explore the integration of this RND-CNN model into clinical diagnostic workflows.