Compared with B-mode imaging, three-dimensional (3D) ultrasound imaging is more helpful in research and clinical application. At present, the 3D ultrasound images can be acquired directly with two-dimensional (2D) array transducer or reconstructed from a series of B-mode images obtained with one-dimensional (1D) array transducer. Imaging with 2D array transducer can achieve a high frame rate, but suffering from the complexity of the imaging system, such as the large amount of channels, and high computational complexity. Reconstructing 3D images from a series of B-mode images can be implemented by recording the position and orientation of the slice images. This is a low-cost and flexible imaging method, but usually suffering from the low imaging quality and low frame rate. In our previous work, a novel 3D ultrasound imaging method in frequency domain with a moved 1D array transducer is presented. This method can reduce the computational complexity with FFT, and get improved imaging quality and frame rate to some extent. Besides, this method can be adopted to construct images with a row–column-addressed 2D array, which can reduce the amount of channels effectively. In this paper, a two-steps implementation of this imaging method is proposed, in which the combined implementation of the 3D imaging is decomposed to two steps of 2D imaging processes in Frequency domain. In the first step, the received echoes of the 1D array transducer at each position are processed with a 2D imaging processes in the lateral- axial planes. In the second step, a 2D imaging processes is preformed in the planes of orthogonal to the transducer. Simulation results show that the two-steps implementation can achieve almost the same imaging quality to the previous work. Compared with the implementation of 3D imaging in our previous work, the proposed two-steps implementation can be carried out with parallel process to improve the computational efficiency, or carried out with loop to reduce the hardware cost. Besides, the first step can be performed with a conventional DAS imaging method when a cylindrical wave is adopted for imaging. The influence of the spread angle of the field is also discussed.

与B模式成像相比，三维（3D）超声成像在研究和临床应用中更有帮助。目前，可以使用二维（2D）阵列换能器直接获取3D超声图像，或从使用一维（1D）阵列换能器获得的一系列B模式图像中重建3D超声图像。使用2D阵列换能器进行成像可以实现高帧频，但是会遭受成像系统的复杂性（例如，大量的通道）和高计算复杂性的困扰。通过记录切片图像的位置和方向，可以实现从一系列B模式图像重建3D图像。这是一种低成本且灵活的成像方法，但通常会遇到成像质量低和帧频低的问题。在我们以前的工作中，提出了一种新颖的3D超声成像方法，该方法在频域上具有移动的1D阵列换能器。该方法可以降低FFT的计算复杂度，并在一定程度上提高成像质量和帧频。此外，该方法可以采用行列寻址的2D数组构造图像，从而可以有效地减少通道数量。在本文中，提出了这种成像方法的两步实现，其中将3D成像的组合实现分解为频域中的2D成像过程的两步。在第一步中，在横向位置上用2D成像过程处理在每个位置接收到的1D阵列换能器的回波。在第二步中，在与换能器正交的平面中执行2D成像过程。仿真结果表明，分两步实施可以实现与先前工作几乎相同的成像质量。与我们之前的工作中的3D成像实现相比，建议的两步实现可以通过并行处理来提高计算效率，或者通过循环执行以降低硬件成本。此外，当采用圆柱波进行成像时，可以使用常规的DAS成像方法执行第一步。还讨论了场扩展角的影响。或通过循环执行以降低硬件成本。此外，当采用圆柱波进行成像时，可以使用常规的DAS成像方法执行第一步。还讨论了场扩展角的影响。或通过循环执行以降低硬件成本。此外，当采用圆柱波进行成像时，可以使用常规的DAS成像方法执行第一步。还讨论了场扩展角的影响。