Development of Real-Time Three-Dimensional Ultrasonic Myocardial Strain Imaging

缺血性心脏病早期常出现由组织刚度改变引起的心肌运动减少。由于心肌表现出复杂的三维运动，为了准确诊断，需要对其刚度分布进行三维评估。然而，传统的组织多普勒和应变率成像技术不能完全满足上述要求，因为它们依赖于角度和平面（二维）处理，并且在大的心肌运动中存在混叠问题。为了克服这些问题，我们提出了利用小孔径二维相控阵跟踪三维运动和基于三维运动分析获得的全应变张量评估心肌功能障碍的新方法。作为一种新的三维心肌运动跟踪方法，我们结合了基于二维相控阵上每个单元的相移量实时测量三维位移矢量的相梯度法和能够精确测量每个单元的大相移量而不出现混叠的组合自相关法。作为一种评估心肌组织刚度的新方法，将测量的三维向量得到的全应变张量不变量可视化为应变不变量图像。由于应变张量不变量与坐标系、视轴和光束方向无关，因此可以独立于入射角评估有效心肌刚度。通过对三维椭圆型心肌模型左心室短轴成像的数值模拟，对所提方法的可行性进行了评价。该模型包括位于1点至3点之间的硬梗死，传统的组织多普勒和应变率成像难以检测到。通过对模型中的PSF和散射系数进行卷积，忠实地模拟了二维阵列探头上各单元接收到的射频信号。采用该方法对模拟回波进行处理后，得到的全应变张量不变图像清晰地描绘了常规成像无法描绘的硬梗死区域。少

Decreases of myocardial motion caused by changes in tissue stiffness often appear in the early stage of ischemic heart disease. Since the myocardium exhibits complex 3-D motion, 3-D assessment of the stiffness distribution is required for accurate diagnosis. However, conventional tissue Doppler and strain-rate imaging techniques cannot meet the above requirement completely because they are angle-dependent and in-plane (2-D) processing, and have the aliasing issue for large myocardial motion.In order to overcome these problems, we propose novel methods to track the 3-D motion by using a 2-D phased array with a small aperture and to assess myocardial malfunction based on full strain tensors obtained by 3-D motion analysis. As a new method of 3-D myocardial motion tracking, we incorporated our phase-gradient method, which is capable of real-time 3-D displacement vector measurement based on the phase shifts obtained at each element on a 2-D phased array, and our combined autocorrelation me … More thod, which accurately measures large phase shifts at each element without aliasing. As a new method of assessing myocardial tissue stiffness, the full strain tensor invariant obtained by the measured 3-D vectors is visualized as a strain invariant image. Since strain tensor invariants are independent of the coordinate system, visual axis and beam direction, valid myocardial stiffness can be assessed independently of incident angle.We evaluated the feasibilities of the proposed methods by numerically simulating the left ventricle short-axis imaging of a 3-D elliptic myocardial model. This model includes a hard infarction, located between 1 and 3 o'clock, where it is difficult to detect by, conventional tissue Doppler and strain-rate imaging. RF signals received at each element on the 2-D array probe were faithfully simulated by convoluting the PSF and the scatter coefficients in the model. When the simulated echoes were processed by the proposed methods, the invariant image obtained by the full strain tensor clearly depicted the hard infarction area where conventional imaging could not. Less