Analysis of the effects of measurement errors on the evaluation of propagation coefficients, in rubber tubes and canine aorta in vivo.

Published measurements of canine arterial propagation coefficients show large disparities in pulse wave attenuation and considerable variation with frequency, suggesting both random and systematic error. Here we devise methods of assessing the likely magnitude of each measurement error source, and calculate the individual and cumulative effects on the propagation coefficients derived using three different schemes to compensate for reflections: the three-point method (3PT), the total occlusion method (OCC) and the general iterative method applied to Cox's arrangement of four transducers (GEN). Results from measurements using impulse excitation in a latex tube show that each method accumulates error in a characteristic frequency pattern. The predicted error bands for OCC are very small except at the fundamental, and clearly exclude a significant further error component. The predicted 3PT error is large at low and at high frequency, and the error bands are better predictors of actual error than in the OCC case. Transducer position error is less than any one of the three pressure measurement errors. The predicted GEN error increases with frequency and when averaged across frequency is of magnitude intermediate between those for OCC and 3PT. Results for the OCC and GEN methods from the canine aorta in vivo generally confirm these trends, but the concentration of natural pulse wave energy at low frequencies acts to reduce low-frequency error and increase high-frequency error.