A new three-component signal model to objectively select power Doppler wall filter cut-off velocity for quantitative microvascular imaging

The wall-filter selection curve (WFSC) method was developed to automatically select cut-off velocities for high-frequency power Doppler imaging. Selection curves are constructed by plotting color pixel density (CPD) as a function of wall filter cut-off velocity. A new three-component mathematical model is developed to guide the design of an online implementation of the method for in vivo imaging. The model treats Doppler imaging as a signal detection task in which the scanner must distinguish intravascular pixels from perivascular and extravascular pixels and includes a cost function to identify the optimum cut-off velocity that provides accurate vascular quantification and minimizes the effect of color pixel artifacts on visualization of vascular structures. The goodness of fit of the three-component model to flow-phantom data is significantly improved compared to a previous two-component model (F test, p < 0:005). Simulations using the new model indicate that selection curves should be sampled using at least 100 cut-off velocities to ensure robust performance of the automated WFSC method and determine an upper bound on CPD variability that ensures reliable vascular quantification accuracy, defined as CPD within 5% of the reference vascular volume fraction. Results of the simulations also provide evidence that limiting the selection of the cut-off velocity to a binary choice between the middle and right end of the characteristic interval is sufficient to meet the quantification accuracy goal. The model provides an intuitive, empirical description of the relationship between system settings and blood-flow detection performance in power Doppler imaging.