基于Hankel-SVD的非平稳超声血流成像杂波抑制技术研究

发布时间：2018-11-19 08:51

【摘要】：有效抑制由血管或血管周围组织时变运动引起的非平稳杂波对于提高诊断超声彩色血流成像中血流动力学参数描述的准确性有着极其重要的意义。该文基于奇异值滤波技术提出一种改进的非平稳杂波自适应抑制方法。该方法逐次利用单个慢时多普勒回波采样矢量构建Hankel矩阵,然后根据奇异值分解后得到的正交Hankel主成份所代表的频域内容,动态选取高阶Hankel主成份重构多普勒血流信号,实现非平稳杂波的有效抑制。为验证算法的有效性,分别对多普勒回波仿真模型合成数据与利用彩色超声设备(Sonix RP)采集的颈动脉血流基带回波信号进行滤波处理,然后采用滞一自相关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,相对于传统IIR滤波方法与多项式回归滤波技术,利用该文所提算法可对高强度、非平稳杂波进行充分抑制,提高血流估计精度,此外,该算法具有空间自适应性,无需人为设定阈值参数以估计杂波空间维数,与现有基于特征分解的自适应滤波方法相比,可以有效提高组织空间高强度时变运动时血流与组织的区分能力。
[Abstract]:It is very important to suppress the nonstationary clutter caused by the time-varying movement of blood vessels or perivascular tissues in order to improve the accuracy of hemodynamic parameters in color hemodynamic imaging. In this paper, an improved adaptive suppression method for nonstationary clutter based on singular value filtering is proposed. In this method, the Hankel matrix is constructed by using a single slow time Doppler echo sampling vector, and then the Doppler blood flow signal is reconstructed by selecting the high order Hankel principal component dynamically according to the frequency domain content represented by the orthogonal Hankel principal component obtained by singular value decomposition. The effective suppression of nonstationary clutter is realized. In order to verify the validity of the algorithm, the synthetic data of Doppler echo simulation model and the baseband echo signal of carotid artery blood flow collected by color ultrasound equipment (Sonix RP) were filtered and processed respectively. Then the average velocity and power of blood flow were calculated by hysteretic-autocorrelation estimation and the imaging was carried out. The processing results show that compared with the traditional IIR filtering method and polynomial regression filtering technique, the proposed algorithm can fully suppress the high intensity, non-stationary clutter and improve the accuracy of blood flow estimation. The algorithm is spatially adaptive and does not require artificial threshold parameters to estimate the spatial dimension of clutter. Compared with the existing adaptive filtering methods based on feature decomposition, the proposed algorithm can be used to estimate the spatial dimension of clutter. It can effectively improve the ability of distinguishing blood flow from tissue during high intensity time-varying movement in tissue space.
【作者单位】：电子科技大学自动化工程学院;国家空气动力研究与发展中心;
【分类号】：R445.1;TN911.7

【参考文献】