BEMD-3DCNN-based method for COVID-19 detection.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Deep Learning

Background:

• The COVID-19 pandemic necessitates rapid and accurate diagnostic tools.
• Existing deep learning models for COVID-19 detection using medical images often overlook spatial and contextual information.
• There is a need for improved detection systems to control the spread of COVID-19.

Purpose of the Study:

• To develop and evaluate a novel 3D Convolutional Neural Network (3DCNN) model for enhanced COVID-19 detection from X-ray images.
• To improve the analysis of spatial and contextual information in medical images for more robust COVID-19 classification.
• To leverage the Bi-dimensional Empirical Mode Decomposition (BEMD) technique for effective image preprocessing.

Main Methods:

• A 3DCNN model was developed, featuring a 3D VGG-16 backbone and a Context-aware Attention (CAA) module.
• Chest X-ray images were preprocessed using Bi-dimensional Empirical Mode Decomposition (BEMD) to create image-based videos.
• The 3DCNN model processed these videos to classify COVID-19, normal, and pneumonia cases.

Main Results:

• The proposed 3DCNN model, incorporating the CAA module, demonstrated effective classification of COVID-19 from X-ray images.
• The model achieved desired results on a dataset comprising 6484 X-ray images (1802 COVID-19 positive).
• The integration of contextual information processing via CAA networks led to superior performance.

Conclusions:

• The developed 3DCNN model with CAA effectively detects COVID-19 by analyzing spatial and contextual features in X-ray images.
• This approach offers a promising advancement in rapid diagnostic systems for infectious diseases.
• The study highlights the potential of 3D deep learning and attention mechanisms in medical image analysis.