Shape Analysis of Coronary Flow Waveforms using Singular Value Decomposition

Coronary flow waveforms exhibit a unique diastolic-dominant structure, characterized by low or retrograde flow during systole. The source material highlights the limited research on the variability of these waveforms and their potential implications for cardiac function. By utilizing Fourier transforms and Singular Value Decomposition (SVD) on baseline data obtained from 32 pigs, the study aimed to investigate this variability.

The pigs represented two breeds (Ossabaw and Yorkshire) and three experimental conditions (lean-control, lean-paced, and obese-paced). The Fourier transforms decomposed the waveforms into 15 harmonics for each pig. Subsequently, an SVD analysis was employed to uncover temporal patterns within these waveforms. The correlations between pig-specific coefficients for the SVD modes and clinical metrics revealed that the second SVD mode shows a significant correlation with heart rate, while the third SVD mode exhibits a strong correlation with mean blood pressure and maximum hyperemic flow.

Additionally, the fourth SVD mode had a weak correlation with left-ventricular end diastolic pressure and endocardial-epicardial flow ratios. This research demonstrates that LAD flow waveforms can be dissected into temporal patterns, which correlate with various physiological features. Moreover, the shape-analysis method employed in this study enables waveform reconstruction and facilitates the visualization of the temporal patterns identified through SVD, offering a distinct advantage over existing methods that primarily characterize flow waveforms based on specific points of interest.