A comparative analysis of coronary and aortic flow waveforms.

Empirical Mode decomposition (EMD) is a mathematical tool designed to analyze non-stationary, non-linear stochastic waves. EMD separates a waveform into its constituent modes of oscillations or intrinsic mode functions (IMFs) and provides meaningful definitions of instantaneous frequency, instantaneous energy, mean trends and oscillation about the mean trends. This study provides a detailed mathematical analysis of blood flow waveforms in the porcine left anterior descending artery and aorta using EMD. Flow data with non-stationary and non-linear characteristics were obtained for several hours using an implanted wireless biotelemetry device. EMD was validated against modern numerical techniques of principal component analysis (PCA) and wavelet analysis by comparing their predicted mean trends and energy distribution. EMD has an advantage over both techniques since it combines the strengths of both: it is adaptive (similar to PCA), and it can define instantaneous frequencies (similar to wavelet analysis). Because of the iterative nature, however, calculations using EMD can be computationally intensive. Sampling rate reduction was used to reduce computation time, without significantly effecting accuracy of IMF calculations. It was found that IMFs calculated at a sampling rate as low as 20 Hz were not significantly different (<6%) from those obtained at the original sampling rate (200 Hz). Our findings suggest that EMD may be a powerful mathematical tool to characterize flow waveforms.