An Improved Convolutional Neural Network Based Approach for Automated Heartbeat Classification.

Area of Science:

• Cardiology
• Biomedical Engineering
• Artificial Intelligence

Background:

• Cardiovascular diseases are linked to aging blood vessels, impacting heart function.
• Electrocardiogram (ECG) is crucial for diagnosing heart disease by recording cardiac electrical activity.
• Arrhythmia detection is complex, necessitating advanced diagnostic tools.

Purpose of the Study:

• To develop an improved convolutional neural network (CNN) model for accurate automatic classification of heartbeats in arrhythmia detection.
• To leverage CNN's automatic feature extraction capabilities for enhanced ECG analysis.
• To validate the proposed CNN model's performance against established standards and databases.

Main Methods:

• Segmentation of individual heartbeats from original ECG signals.
• Implementation of a CNN with convolutional layers utilizing kernels of different sizes for multi-scale feature extraction.
• Application of max-pooling and fully-connected layers for classification.
• Experimentation adhering to the AAMI inter-patient standard, classifying normal (N), supraventricular ectopic (S), ventricular ectopic (V), fusion (F), and unknown (Q) beats.
• Validation using the MIT arrhythmia database.

Main Results:

• The proposed improved CNN model automatically classifies different types of arrhythmia with high accuracy.
• Achieved an accuracy of 99.06% in arrhythmia detection.
• Demonstrated superior performance compared to traditional machine learning methods by eliminating manual feature extraction.

Conclusions:

• The developed improved CNN model is effective for automatic arrhythmia detection from ECG.
• The model's ability to process features at different scales contributes to its high accuracy.
• This CNN model shows potential as a clinical tool for automated diagnosis of cardiac arrhythmias.