基于伽马混合模型B超图像聚类配准的颈动脉管壁搏动位移干扰抑制

Carotid artery wall pulsation displacement interference suppression based on ultrasonic images clustered registration with the Gamma mixture model

提出了基于伽马混合模型B超图像聚类(UICG)的配准方法以抑制血管壁搏动位移估计的干扰。使用伽马混合模型对颈动脉B超图像聚类,以远距组织的归一化互信息为相似性测度提取干扰曲线,基于干扰曲线对聚类图像序列进行空间逆变换以消除外界干扰,最后采用散斑追踪法在配准后的聚类图像序列中估计管壁搏动位移。仿真实验表明,相比主轴质心与互信息相结合(PCMI)的传统配准算法,X、Y和旋转方向上的UICG干扰归一化均方根误差(NRMSE)分别减小了36%、38%和32%,管壁搏动位移估计的NRMSE均值减小了37%。对健康受试者颈动脉的实测试验进一步证明了UICG法的有效性。综上,UICG法能有效抑制干扰,提高管壁搏动位移的估计精度。

A registration method based on ultrasound images clustering with the Gamma mixture model(UICG) is proposed to suppress the interference in the wall pulsation displacement estimation. A Gamma mixture model is used for the carotid artery B-mode ultrasound image clustering. Then, the normalized mutual information of the distant tissues is used as the similarity measure to extract interference curves. Based on the interference curves, the spatial inverse transformation of the clustering image sequence is implemented to eliminate the external interference. Finally, with the speckle tracking method, the pulsation displacement of the arterial wall is estimated from the sequence of the registered cluster images. Compared with the traditional registration algorithm, namely, the position weighted principal axis and centroid method combined with mutual information(PCMI), simulations show that the normalized root mean square errors(NRMSEs) of the UICG-based interference estimations in X, Y, and rotation directions are reduced by 36%, 38%, and 32%, respectively. The mean NRMSE of the estimated wall displacements is decreased by 37%. The clinical trials based on the common carotid arteries of healthy subjects further evaluate the effectiveness of the UICG method. In summary, the UICG method can effectively suppress interference and thus improve the measurement accuracy of the wall displacement.