An improved parametric relaxation approach to blood flow signal estimation with single-ensemble samples in color flow imaging

In color flow imaging, it is a challenging work to accurately extract blood flow information from ultrasound Doppler echoes dominated by strong clutter components. Conventional non-parametric estimators usually cause flow velocity estimation biases since clutter rejection filters often distort parts of blood flows or fail to suppress clutter adequately. In this paper, a parametric estimation framework called relaxation (RELAX) is proposed to directly extract blood flow information from raw ultrasound Doppler signals. RELAX constructs an exponential model to approximate single-ensemble ultrasound Doppler echoes and solves for its parameters in a decoupled manner. The principal Doppler frequencies of the clutter and the blood flow obtained by RELAX are independent of the corresponding phase shifts. A parameter selection algorithm based on the energy ratio is proposed to determine the number of principal components. A series of simulations shows that the proposed RELAX approach can achieve accurate velocity estimation of blood flow. The mean overall errors obtained by RELAX are 30% lower compared to those obtained using state-of-the-art non-parametric methods using eigen-decomposition based filters. RELAX also eliminates the effect caused by white noise and achieves an extremely low estimation variance of low-velocity blood flow ( < 15cm/s) compared to those obtained using competing methods. Clinical experiment results show that the RELAX method leads to the highest blood flow energy and blood-to-clutter energy ratio among those obtained using the discussed methods.