Computational method for high resolution spectral analysis of fractionated atrial electrograms

Area of Science:

• Biomedical Engineering
• Signal Processing
• Cardiology

Background:

• Discrete Fourier Transform (DFT) is commonly used for spectral analysis of complex fractionated atrial electrograms (CFAE) in atrial fibrillation (AF).
• DFT's time resolution is limited by its inverse relationship with frequency resolution, often proving unsatisfactory for dynamic electrogram analysis.
• A need exists for spectral estimators with enhanced time-frequency resolution for CFAE analysis.

Purpose of the Study:

• To compare the performance of a novel spectral estimator (NSE) against the traditional DFT for CFAE analysis.
• To evaluate the NSE's ability to provide improved time-frequency resolution for spectral parameter computation.
• To determine if NSE enhances the discrimination of spectral differences between paroxysmal and persistent AF.

Main Methods:

• A novel spectral estimator (NSE) based on signal averaging was computationally implemented and compared with DFT.
• Both estimators model the autocorrelation function to derive the power spectrum, but NSE's resolution is rate/period(2)-dependent, not window-length dependent.
• Performance was assessed by analyzing frequency detection error in a simulated electrogram with phase jitter and by comparing spectral parameter differences (dominant amplitude, dominant frequency, mean spectral profile) between paroxysmal and persistent AF using varying window lengths (0.5s to 8s).

Main Results:

• NSE demonstrated significantly lower frequency detection error compared to DFT across all time resolutions (p<0.001).
• NSE achieved 0.5s time resolution with 0.05 Hz frequency resolution, outperforming DFT's 2s time resolution with 0.5 Hz frequency resolution.
• NSE revealed greater significant differences in dominant amplitude (p<0.0001) and improved discrimination for dominant frequency and mean spectral profile between AF types compared to DFT.

Conclusions:

• The novel spectral estimator (NSE) exhibits superior performance compared to the discrete Fourier transform (DFT) for measuring spectral properties of CFAE.
• NSE offers enhanced time-frequency resolution, leading to more accurate and discriminative analysis of atrial electrograms in AF.
• These findings suggest NSE as a valuable tool for advancing the understanding and diagnosis of atrial fibrillation subtypes.