Aiming at rejecting nonstationary clutter originating from slow moving vessels and surrounding tissues,a singular-spectral-weighting-based clutter suppression method for color ultrasound Doppler imaging is presented in this paper. First,a Hankel data matrix is created from each slow-time ensemble.Then,the singular value decomposition is performed to obtain the orthogonal basis functions for regression filtering.The coefficients of the filter are adaptively computed by a modi-fied sigmoid function from the power normalized singular spectral,which allows for means to detect regions of clutter arti-facts with high specificity.To analyze the efficacy of the proposed adaptive filter,in-vitro experiments were carried out.For the experiments,raw CFI data were acquired by a commercial ultrasound system (Sonix RP,Ultrasonix Inc.).Then,blood flow parameters are estimated from an estimate of the lag-one auto correlator of the filtered signal.The reconstructed flow and power images verifies that the proposed method outperforms other tested methods in rejecting high intense nonstationary clutter,which leads to improved distinguishing between blood and tissue regions.%针对超声彩色多普勒成像中由血管或血管周围组织时变运动引起的非平稳杂波抑制问题，提出一种基于奇异值谱加权的杂波抑制算法．首先根据单个慢时方向的回波多普勒矢量构建Hankel数据矩阵并进行奇异值分解，利用分解得到的Hankel主成分构造回归滤波器的正交基函数，同时引入改进的Sigmoid函数根据能量归一化奇异值谱计算回归滤波器系数，使得杂波区域的检测具有高度的特异性，从而提高非平稳杂波的抑制能力．为验证算法的有效性，利用商业级超声仪（Sonix RP，Ultrasonix Inc．）采集50帧人体颈动脉血流基带回波数据并进行滤波处理，滤波后数据采用滞一自相关法估计法计算血流平均速度与功率并进行成像．处理结果表明，与传统静态滤波器以及现有基于特征值分解的滤波算法相比，可有效增强组织空间高强度时变运动时血流与组织的区分能力．
展开▼
