COV-ECGNET: COVID-19 detection using ECG trace images with deep convolutional neural network

Area of Science:

• Artificial Intelligence
• Cardiology
• Medical Imaging

Background:

• Accurate and swift COVID-19 detection is vital for public health.
• Existing detection methods vary in accessibility and speed.
• Electrocardiogram (ECG) data offers potential for non-invasive disease identification.

Purpose of the Study:

• To investigate the efficacy of deep convolutional neural network (CNN) models for COVID-19 detection using ECG trace images.
• To compare the performance of various CNN architectures in classifying normal, COVID-19, and other cardiovascular diseases (CVDs).
• To explore the potential of ECG-based AI for accessible, low-resource diagnostics.

Main Methods:

• Utilized a public dataset of 1937 ECG images across five categories: normal, COVID-19, myocardial infarction (MI), abnormal heartbeat (AHB), and recovered MI (RMI).
• Employed six deep CNN models (ResNet18, ResNet50, ResNet101, InceptionV3, DenseNet201, MobileNetv2).
• Evaluated three classification schemes: two-class (normal vs. COVID-19), three-class (normal, COVID-19, other CVDs), and five-class (all categories).

Main Results:

• DenseNet201 achieved 99.1% accuracy for two-class and 97.36% for three-class classification.
• InceptionV3 demonstrated 97.83% accuracy for the five-class classification.
• ScoreCAM visualization confirmed model focus on relevant ECG image features.

Conclusions:

• Deep CNN models can effectively detect COVID-19 from ECG images.
• The proposed method shows promise for rapid, computer-aided diagnosis of COVID-19 and cardiac conditions, especially in resource-limited settings.
• Smartphone-captured ECGs could facilitate widespread screening.