ECG fiducial point extraction using switching Kalman filter.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Computational Biology

Background:

• Accurate fiducial point (FP) extraction from electrocardiogram (ECG) signals is crucial for diagnosing cardiac conditions.
• Existing methods often struggle with baseline wander and noise, impacting precision.
• Modeling ECG beats and baselines with mathematical functions is a common approach.

Purpose of the Study:

• To propose a novel method for extracting ECG fiducial points (FPs) using a switching Kalman filter (SKF).
• To model ECG waveforms and baselines using Gaussian and autoregressive models, respectively.
• To evaluate the performance of the proposed SKF method against existing techniques.

Main Methods:

• Utilized a switching Kalman filter (SKF) with a discrete state variable ('switch') operating across 7 modes.
• Modeled ECG beats (P-wave, QRS complex, T-wave) with Gaussian functions and baselines with first-order autoregressive models.
• Estimated an ECG signal path based on mode probabilities to identify FPs.

Main Results:

• The proposed SKF method achieved a mean error of 2 ms (less than one sample) and a root mean square error (RMSE) of 14 ms.
• These error metrics were significantly lower compared to wavelet transform, partially collapsed Gibbs sampler (PCGS), and extended Kalman filter methods.
• The SKF method demonstrated reduced RMSE and smaller variability in FP extraction.

Conclusions:

• The novel switching Kalman filter (SKF) method offers superior accuracy for fiducial point extraction in ECG signals.
• This approach provides a robust and precise tool for analyzing cardiac electrophysiological data.
• The method's performance validates its potential for clinical applications in ECG interpretation.