Extraction and characterisation of underlying velocity waveforms in poststenotic flow

The problem of extracting and characterising underlying velocity waveforms in poststenotic flow is examined using a set of hot-film anemometer measurements in the descending thoracic aorta of a dog. The question of separating turbulence from the underlying waveform is studied by applying ensemble averaging and lowpass filtering methods. Results conclusively demonstrate that lowpass filtering is inadequate for defining the underlying waveform, and thus ensemble averaging is required. The effect of the number of beats taken in forming the ensemble averaging is required.The effect of the number of beats taken in forming the ensemble is examined with the resultant recommendation that at least 20 cycles be employed. Finally, the use of autoregressive methods of spectral analysis are introduced to determine the dominant frequencies in the underlying waveform; and these are shown to be superior to Fourier transform methods for this purpose, provided note is taken of appropriate relationships between the sampling frequency and the autoregressive model order.