In pulsed Doppler ultrasound systems, the ultrasound radiofrequency (RF) signals received can be employed to estimate noninvasively the time-dependent blood flow velocity distribution within an artery. The RF signals are composed of signals originating from clutter (e.g., vessel walls) and scatterers (e.g., red blood cells). The clutter, which is induced by stationary or slowly-moving structure interfaces, must be suppressed to get reliable estimates of the mean blood flow velocities. In conventional pulsed Doppler systems, this is achieved with a static temporal high-pass filter. The static cut-off frequency and the roll-off of these filters cause the clutter not always to be optimally suppressed. This paper introduces a clutter removal filter that is based on Singular Value Decomposition (SVD). Unlike conventional high-pass filters, which take into account only the information of the temporal direction, the SVD filter makes use of the information of the temporal and spatial directions. The advantage of this approach is that it does not matter where the clutter is located in the RF signal. The performance of the SVD filter is examined with computer-generated Doppler RF signals. The results are compared with those of a standard linear regression (SLR) filter. The performance of the SVD filter is good, especially if a large temporal window (i.e., approximately 100 RF signals) is applied, which improves the performance for low blood flow velocities. A major disadvantage of the SVD filter is its computational complexity, which increases considerably for larger temporal windows.
Cerebral blood flow measurement by dynamic contrast MRI using singular value decomposition with an adaptive threshold
