A wavelet-based ECG delineator: evaluation on standard databases.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Cardiology

Background:

• Accurate electrocardiogram (ECG) delineation is crucial for diagnosing cardiac conditions.
• Existing automated ECG analysis methods often struggle with precise wave delineation.

Purpose of the Study:

• To develop and evaluate a robust single-lead ECG delineation system using the wavelet transform (WT).
• To accurately detect QRS complexes and delineate P, T waves, and their components.

Main Methods:

• Utilized the wavelet transform (WT) for robust ECG signal processing.
• Implemented a multi-step approach for QRS complex detection and subsequent wave delineation (peaks, onset, end).
• Validated the algorithm on multiple established, manually annotated ECG databases (MIT-BIH Arrhythmia, QT, European ST-T, CSE).

Main Results:

• Achieved high sensitivity (Se = 99.66%) and positive predictivity (P+ = 99.56%) for QRS detection across validation databases (>980,000 beats).
• Attained over 99.8% Se and P+ on the MIT-BIH Arrhythmia Database.
• Delineation errors (mean difference) were within one sampling interval, with standard deviations comparable to inter-physician variability.
• Demonstrated superior performance compared to other algorithms, particularly in T-wave end determination.

Conclusions:

• The wavelet transform-based ECG delineation system is robust and accurate.
• The system demonstrates high performance in QRS detection and wave delineation, meeting clinical validation standards.
• This WT approach offers an improved solution for automated ECG analysis.