Respiration effect on wavelet-based ECG T-wave end delineation strategies

Area of Science:

• Cardiology
• Biomedical Engineering
• Signal Processing

Background:

• Accurate T-wave end delineation is crucial for electrocardiogram (ECG) analysis.
• Respiration-induced mechanical effects can introduce variability in T-wave end location.
• Existing automated delineation methods may be susceptible to these respiratory artifacts.

Purpose of the Study:

• To investigate the impact of respiration on T-wave end delineation using wavelet transform (WT).
• To compare the performance of single-lead (SL), single-lead referenced (SLR), and multilead (ML) delineation approaches.
• To quantify the relationship between respiratory variations and T-wave end location errors.

Main Methods:

• Utilized wavelet transform (WT) for automatic T-wave end delineation.
• Compared SL, SLR, and ML delineation strategies on simulated and real ECG data.
• Employed spectral coherence and ARARX modeling to quantify the influence of respiration.
• Analyzed the vectorcardiographic spatial loop's T-wave end region in relation to respiration.

Main Results:

• Multilead (ML) delineation reduced T-wave end location error dispersion by approximately 40% compared to single-lead (SL).
• The percentage of T-wave end error correlated with respiration decreased from 99% (SL) to 72% (ML).
• Single-lead referenced (SLR) showed intermediate improvements over SL.

Conclusions:

• Multilead (ML) approaches are superior for T-wave end delineation in the presence of respiratory motion.
• ML methods significantly reduce the instability and error in T-wave end location caused by respiration.
• The findings support the adoption of ML strategies for more robust ECG analysis.