A robust method for ECG-based estimation of the respiratory frequency during stress testing

Area of Science:

• Cardiology
• Biomedical Engineering
• Signal Processing

Background:

• Electrocardiogram (ECG) signals during stress testing are often corrupted by nonstationary noise and QRS morphology changes.
• Existing respiratory frequency estimation methods struggle with the dynamic and noisy nature of exercise ECGs.

Purpose of the Study:

• To develop a robust method for estimating respiratory frequency from ECG signals during stress testing.
• To address the limitations of current methods in handling noisy and nonstationary ECG data.

Main Methods:

• The method utilizes the oscillatory pattern of the heart's electrical axis rotation angles, induced by respiration.
• Rotation angles are derived from least-squares loop alignment, followed by power spectral analysis for frequency estimation.
• Robust techniques are incorporated to manage the nonstationary characteristics of exercise ECGs.

Main Results:

• For simulated signals, the respiratory frequency estimation error was 0.5% +/- 0.2% (0.002 +/- 0.001 Hz).
• The error between ECG-derived and airflow-measured respiratory frequencies was 5.9% +/- 4% (0.022 +/- 0.016 Hz) in human subjects.
• The method demonstrated high accuracy and suitability for analyzing noisy ECGs during stress tests.

Conclusions:

• The presented method offers a robust approach for respiratory frequency estimation from ECG during stress testing.
• It effectively handles the challenges posed by nonstationary noise and signal variations in exercise ECGs.
• This technique shows significant potential for improved patient monitoring and analysis in clinical settings.