Atrial activity extraction from single lead ECG recordings: evaluation of two novel methods

Area of Science:

• Biomedical Engineering
• Cardiovascular Physiology
• Signal Processing

Background:

• Atrial fibrillation (AF) diagnosis relies on identifying atrial activity (AA) in electrocardiograms (ECGs).
• Extracting the faint AA signal from noisy single-lead ECGs, particularly during AF, remains a challenge.
• Existing methods like averaged beat subtraction (ABS) have limitations in accuracy.

Purpose of the Study:

• To develop and evaluate novel algorithms for enhanced atrial activity (AA) signal extraction from single-lead electrocardiograms (ECGs).
• To compare the performance of proposed methods against existing techniques for AF analysis.

Main Methods:

• Weighted Average Beat Subtraction (WABS): Constructing a QRS template by minimizing mean square error for beat subtraction.
• Maximum Likelihood Estimation (MLE): Estimating probability density functions using a generalized Gaussian model for signal separation.
• Performance Evaluation: Utilizing simulated and clinical ECG data to assess algorithm accuracy.

Main Results:

• The WABS method demonstrated a 23.5% reduction in normal mean square error compared to ABS.
• The MLE-based algorithm achieved a 20.2% reduction in normal mean square error compared to ABS.
• Both proposed methods showed improved performance in isolating the AA signal.

Conclusions:

• WABS and MLE are effective and improved methods for extracting atrial activity signals from single-lead ECGs in atrial fibrillation.
• These advanced signal processing techniques offer potential for more accurate AF diagnosis and monitoring.
• Further validation on larger clinical datasets is warranted to confirm clinical utility.