Improved alignment method for noisy high-resolution ECG and Holter records using multiscale cross-correlation

Area of Science:

• Biomedical Engineering
• Signal Processing
• Cardiology

Background:

• Accurate fiducial point estimation is crucial for coherent signal averaging in high-resolution electrocardiogram (HRECG) analysis.
• Traditional temporal cross-correlation alignment methods struggle with precise fiducial marking in noisy HRECG records, particularly from Holter systems.

Purpose of the Study:

• To introduce and evaluate a novel multiscale cross-correlation alignment method for HRECG signal processing.
• To assess the performance of the proposed method against traditional temporal methods in noisy HRECG data.

Main Methods:

• Developed a multiscale cross-correlation technique for aligning detected beats to a template beat.
• Tested the multiscale and temporal alignment methods on simulated HRECG records with various noise types (white, electromyographic, power line interference) and levels.
• Validated the methods using simulated records derived from real Holter HRECG data.

Main Results:

• The multiscale alignment method consistently demonstrated lower trigger jitter compared to the temporal method across all tested noise conditions.
• The proposed method proved effective even with significant levels of simulated noise, including white noise, electromyographic noise, and 50 Hz power line interference.
• Performance was validated on simulated records constructed from actual noisy Holter HRECG data, confirming its practical applicability.

Conclusions:

• The proposed multiscale cross-correlation alignment method offers superior performance over traditional temporal methods for HRECG signal averaging.
• This novel approach effectively mitigates the challenges posed by high noise levels in HRECG recordings, such as those encountered in Holter monitoring.
• The multiscale method provides a more precise fiducial mark, leading to improved signal averaging and potentially more accurate diagnostic interpretations from noisy HRECG data.