From ECG signals to images: a transformation based approach for deep learning

Area of Science:

• Cardiology
• Biomedical Engineering
• Artificial Intelligence

Background:

• Ventricular arrhythmias (VA) are irregular heart rhythms linked to heart disease.
• Electrocardiogram (ECG) signals are crucial for diagnosis but are complex and nonlinear.
• Current analysis methods for ECG signals can be challenging.

Purpose of the Study:

• To develop a novel deep learning approach for accurate VA detection.
• To transform ECG signals into image data for enhanced analysis.
• To improve the diagnostic accuracy of cardiac arrhythmias.

Main Methods:

• ECG signals were converted into images.
• AlexNet, VGG-16, and Inception-v3 deep learning models were trained using transfer learning.
• Feature fusion and selection techniques were employed.
• Supervised learning classifiers were used for final classification.

Main Results:

• The proposed deep learning model achieved 97.6% accuracy in VA detection on the MIT-BIH dataset.
• Cubic Support Vector Machine was utilized as the final classifier.
• The novel image-based ECG analysis demonstrated effectiveness.

Conclusions:

• The deep learning approach offers a promising, highly accurate method for VA detection.
• Transforming ECG signals into images is an effective preprocessing step.
• This method can significantly aid in the diagnosis of cardiac arrhythmias.