Spectral estimation and analysis of LDA data in pulsatile flow through heart valves

The Laser Doppler Anemometry (LDA) measurement technique presents an inherent difficulty when spectral analysis is applied to it. The random nature of the LDA signal prohibits sampling at regular, equi-spaced, time instants. Irregular sampling presents additional variability of the spectral estimator. In order to reduce this variability, spectral analysis of LDA data is performed according to the method of direct Fourier transform of short blocks of data, as suggested by Gaster and Roberts. The LDA data is measured in a flow field distal to prosthetic heart valves with varying degrees of stenosis. The spectral estimates of velocity data sampled during the rapid closure stage of the valve are achieved with excellent frequency resolution. Important and useful information about dominant frequency peaks and preferred modes which exist in the flow, otherwise smeared or concealed in the spectral contents, are then derived from the spectral information. These modes are quantitatively analysed in light of vortex formation and related flow mechanisms. Comparative studies of normal and stenosed valves show that the preferred modes are governed by the valve geometry and dynamic behavior and are correlated to the severity of the stenosis.