颈动脉B超估计血管壁搏动位移的干扰抑制  

A Disturbance Rejection Method for Displacement Estimation of Carotid Wall

作　　者：张晴晖 张榆锋[1] 林文晶[1] 周屹[1] 章克信[3] 高莲[1] 

机构地区：[1]云南大学信息学院电子工程系,云南昆明650091 [2]西南林业大学计算机与信息学院,云南昆明650224 [3]昆明医学院第二附属医院,云南昆明650031

出　　处：《电子学报》2017年第12期2903-2908,共6页Acta Electronica Sinica

基　　金：国家自然科学基金(No.61261007;No.61561049);云南省自然科学基金重点项目(No.2013FA008)

摘　　要：超声检测时,由于测量者把握探头的不稳定、受试者的呼吸运动等造成的干扰,将对通过B超图像进行斑点跟踪提取到的动脉壁位移的精度造成影响.为此,本文提出基于B超序列相位图像的全局刚性特征进行配准消除干扰.相位特征通过对B超图像进行Riesz变换,再采用高斯差分法提取得到.为了更快地完成寻优,减少配准误差,采用了基于位置加权的主轴质心法与互信息相结合的分阶段的配准方法.结果表明,与直接对B超或相位图采用斑点跟踪的方法相比,该方法有效去除了干扰,管壁位移提取的精度有显著提高.本文方法还可用于心脑等医学图像的配准.The disturbances caused by the instability of probe hold and the breathing exercise from subjects, are introduced during B-ultrasound examination. It decreases the measuring accuracy of the arterial wall displacement obtained from the B-Mode images by speckle tracking method. Thus,we propose a global rigid-registration method based on phase images of B-Mode images to eliminate these external disturbances. The phase features are obtained from B-Mode images via Riesz transform and difference of Gaussian filters. To improve the speed of searching optimization and the registration accuracy,a stepwise registration method is applied based on position weighted principal axis and centroid method combined with mutual information. The results show that,compared with these schemes of extracting wall displacement directly from B-Mode or phase images by speckle tracking method, this scheme can effectively remove these disturbances, and significantly improve the estimation accuracy. The proposed method can also be used in image registration of cardio-cerebro diseases.

关 键 词：B型超声 血管壁搏动检测 刚性配准 相位特征 主轴质心法 斑点跟踪 互信息 各向异性斑点减少 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]