Improving Peak Velocity Estimation Accuracy in EchoPIV using Anisotropic Windows

The use of isotropic interrogation windowing functions in echo-particle image velocimetry (echoPIV) introduces a low pass filter (LPF) effect, resulting in underestimation of peak velocities and spatial smoothing of the velocity profiles. Therefore, adaptive anisotropic windowing could be beneficial, especially in regions with high velocity gradients. For this an elliptical windowing function defined by prior estimation of the velocity field is used to replace the standard (Gaussian) window applied in conventional echoPIV processing. The proposed elliptical windows were tested in an in silico ultrasound carotid flow-phantom. The windowing functions were tested in combination with window-refinement. When ending on a relatively course window size, the elliptical windowing functions resulted in a 7.9% improvement in peak velocity accuracy, compared to Gaussian windowing, in the internal carotid during the peak systolic phase. When the window size was refined to the limit of the point spread function (PSF), the anisotropic windowing function no longer achieved improved accuracy compared with Gaussian windowing. Anisotropic windowing is a promising method for improving the detection of peak velocities in jet-like flow profiles.