Detecting COVID-19 in Chest X-Ray Images via MCFF-Net

Area of Science:

• Artificial Intelligence
• Medical Imaging
• Computational Biology

Background:

• COVID-19, a severe respiratory illness caused by SARS-CoV-2, has overwhelmed global health systems.
• Current RT-PCR testing for COVID-19 faces limitations including high cost, time consumption, and variable sensitivity.
• Chest X-ray (CXR) imaging presents unique characteristics that can be leveraged for disease detection.

Purpose of the Study:

• To develop an efficient deep learning model for COVID-19 detection using CXR images.
• To introduce a novel Parallel Channel Attention Feature Fusion Module (PCAF) for enhanced feature extraction.
• To evaluate the performance of the proposed MCFF-Net architecture in classifying COVID-19 cases.

Main Methods:

• Designed the Parallel Channel Attention Feature Fusion Module (PCAF) for CXR analysis.
• Developed a new convolutional neural network, MCFF-Net, incorporating the PCAF module.
• Utilized three distinct classifiers (1-FC, GAP-FC, Conv1-GAP) to optimize recognition efficiency.

Main Results:

• The MCFF-Net66-Conv1-GAP model achieved an overall accuracy of 94.66% for 4-class classification.
• Specific performance metrics for COVID-19 classification were 100% for accuracy, precision, sensitivity, specificity, and F1-score.
• The model demonstrated high efficacy in distinguishing COVID-19 from other conditions.

Conclusions:

• MCFF-Net provides a highly accurate and efficient method for COVID-19 diagnosis via CXR analysis.
• The AI model can assist clinicians in making timely and appropriate diagnostic decisions.
• This approach offers a potential solution to mitigate the impact of limited COVID-19 testing kit availability.