Autoregressive spread-spectral analysis of intracardiac electrograms: comparison to Fourier analysis

The combination of autoregressive (AR) spectral estimation and the spread spectrum is investigated as a method of evaluating cardiac electrogram recordings. The authors provide examples of autoregressive spectral estimation applied to nonfibrillatory and fibrillatory signals and demonstrate its advantages over established spectral analysis algorithms. AR models were fit to brief epochs of atrial fibrillation, ventricular fibrillation, sinus rhythm, and other nonfibrillatory rhythm electrograms. Spread-spectrum plots were constructed from epoch spectra to examine temporal changes in cardiac dynamics as well as differences in spectral estimates. The results of applying the high resolution spread spectrum to clinical data in relation to proposed physiological mechanisms of fibrillation are examined