Developing of robust and high accurate ECG beat classification by combining Gaussian mixtures and wavelets features

Area of Science:

• Cardiology
• Biomedical Engineering
• Signal Processing

Background:

• Electrocardiogram (ECG) beat classification is crucial for computer-aided diagnosis.
• Accurate identification of arrhythmias aids in timely medical intervention.

Purpose of the Study:

• To develop and evaluate a novel method for ECG beat classification.
• To classify ECG beats into six distinct categories, including normal and various arrhythmias.

Main Methods:

• Utilized 10,502 ECG beats from the MIT-BIH Arrhythmia database.
• Extracted features using Gaussian mixture coefficients and wavelets.
• Applied Principal Component Analysis (PCA) for feature reduction.
• Employed Probabilistic Neural Network (PNN) and Random Forest (RF) classifiers.

Main Results:

• Achieved high classification accuracy: 99.99% for PNN and 99.97% for RF.
• Demonstrated excellent precision, sensitivity, and specificity for both classifiers.
• Confirmed the effectiveness of combined Gaussian mixture and wavelet features for arrhythmia classification.

Conclusions:

• The proposed method effectively classifies ECG beats with high accuracy.
• The combination of Gaussian mixture and wavelet features provides valuable insights into heart performance.
• This approach shows significant potential for clinical application in arrhythmia diagnosis.