A deep learning-based framework for detecting COVID-19 patients using chest X-rays.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Infectious Disease Diagnostics

Background:

• Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) has led to global COVID-19 outbreaks.
• Rapid and accurate COVID-19 detection is crucial for epidemic containment.
• Radiography, including chest X-rays, shows promise for early COVID-19 prediction and timely patient management.

Purpose of the Study:

• To develop a highly efficient, lightweight Convolutional Neural Network (CNN) architecture for detecting COVID-19 from chest X-ray images.
• To propose a robust deep learning system for reliable COVID-19 detection.
• To achieve a low false-negative rate in classifying COVID-19 positive cases.

Main Methods:

• Evaluated pre-trained deep learning models (InceptionV3, Xception, MobileNetV2, NasNet, DenseNet201) for medical image classification.
• Developed and proposed a novel, lightweight, shallow CNN architecture.
• Trained and tested the models on a dataset of 2,541 chest X-rays from public databases, including confirmed COVID-19 positive and healthy cases.

Main Results:

• The proposed shallow CNN achieved a maximum accuracy of 99.68%.
• Key performance metrics included sensitivity of 99.66%, specificity of 99.70%, and AUC of 99.98%.
• The developed model demonstrated superior performance with fewer parameters and lower complexity compared to existing state-of-the-art deep learning models.

Conclusions:

• The proposed lightweight CNN is a highly effective tool for rapid and accurate COVID-19 detection from chest X-rays.
• This model can significantly assist healthcare professionals in faster patient screening and improved treatment strategies.
• The findings suggest the model's potential to aid in controlling the spread of the COVID-19 epidemic.