Artifacts in blood velocity estimation using ultrasound and cross-correlation.

Estimation of blood velocities using ultrasound and time-domain cross-correlation is investigated. The measurement principle is introduced, and the basic properties of the cross-correlation function are discussed. Expressions for the variance of the estimates of the peak location are given, showing the influence of integration time, transducer bandwidth and signal-to-noise ratio. It is also shown that the technique can be implemented using only the sign of the signals. A simple simulation program is used in order to study the exact influence from the different parameters. Using the program, it is shown that any velocity can result from the estimation with a certain probability regardless of the true velocity, when the signal-to-noise ratio is low. This is due to the non-linear estimation technique employed and shows that estimation variance yields little information for this estimator. Graphs are given for the probability of correct estimation, when the different parameters are varied.