Arrhythmia detection and classification using morphological and dynamic features of ECG signals

Area of Science:

• Cardiology
• Biomedical Engineering
• Signal Processing

Background:

• Cardiac arrhythmias require accurate detection and classification for effective management.
• Existing computer-assisted methods for arrhythmia analysis have limitations.

Purpose of the Study:

• To develop and validate a novel, highly accurate computer-assisted approach for cardiac arrhythmia classification.
• To combine morphological and dynamic features for improved classification performance.

Main Methods:

• Extraction of morphological features using Wavelet Transform (WT) and Independent Component Analysis (ICA) on individual heartbeats.
• Extraction of dynamic features from RR interval information characterizing heart rhythm.
• Concatenation of morphological and dynamic features, followed by classification using Support Vector Machine (SVM) into 15 classes.
• Fusion of results from two independent ECG lead analyses for enhanced decision-making.

Main Results:

• The proposed method achieved an overall accuracy of 99.66% across 85,945 heartbeats from the MIT-BIH Arrhythmias Database.
• The combined feature approach significantly improved classification performance compared to methods using single feature types.
• The lead fusion strategy enhanced classification robustness and confidence.

Conclusions:

• The novel approach integrating morphological and dynamic features with SVM classification offers superior accuracy for cardiac arrhythmia detection.
• This method represents a significant advancement in computer-assisted cardiac disorder management.
• The high accuracy achieved suggests potential for clinical application in real-time arrhythmia monitoring.