Ma Ren, Zhou Xiaoqing, Zhang Shunqi, Yin Tao, Liu Zhipeng
Institute of Biomedical Engineering, Chinese Academy of Medical Sciences and Peking Union Medical College, Tianjin 300192, People's Republic of China.
Phys Med Biol. 2016 Dec 21;61(24):8762-8778. doi: 10.1088/1361-6560/61/24/8762. Epub 2016 Nov 28.
In this study we present a three-dimensional (3D) reconstruction algorithm for magneto-acoustic tomography with magnetic induction (MAT-MI) based on the characteristics of the ultrasound transducer. The algorithm is investigated to solve the blur problem of the MAT-MI acoustic source image, which is caused by the ultrasound transducer and the scanning geometry. First, we established a transducer model matrix using measured data from the real transducer. With reference to the S-L model used in the computed tomography algorithm, a 3D phantom model of electrical conductivity is set up. Both sphere scanning and cylinder scanning geometries are adopted in the computer simulation. Then, using finite element analysis, the distribution of the eddy current and the acoustic source as well as the acoustic pressure can be obtained with the transducer model matrix. Next, using singular value decomposition, the inverse transducer model matrix together with the reconstruction algorithm are worked out. The acoustic source and the conductivity images are reconstructed using the proposed algorithm. Comparisons between an ideal point transducer and the realistic transducer are made to evaluate the algorithms. Finally, an experiment is performed using a graphite phantom. We found that images of the acoustic source reconstructed using the proposed algorithm are a better match than those using the previous one, the correlation coefficient of sphere scanning geometry is 98.49% and that of cylinder scanning geometry is 94.96%. Comparison between the ideal point transducer and the realistic transducer shows that the correlation coefficients are 90.2% in sphere scanning geometry and 86.35% in cylinder scanning geometry. The reconstruction of the graphite phantom experiment also shows a higher resolution using the proposed algorithm. We conclude that the proposed reconstruction algorithm, which considers the characteristics of the transducer, can obviously improve the resolution of the reconstructed image. This study can be applied to analyse the effect of the position of the transducer and the scanning geometry on imaging. It may provide a more precise method to reconstruct the conductivity distribution in MAT-MI.
在本研究中，我们基于超声换能器的特性，提出了一种用于磁感应磁声层析成像（MAT-MI）的三维（3D）重建算法。对该算法进行了研究，以解决由超声换能器和扫描几何结构引起的MAT-MI声源图像模糊问题。首先，我们使用来自实际换能器的测量数据建立了换能器模型矩阵。参照计算机断层扫描算法中使用的S-L模型，建立了电导率的3D体模模型。在计算机模拟中采用了球形扫描和柱形扫描几何结构。然后，利用有限元分析，通过换能器模型矩阵可以获得涡流、声源以及声压的分布。接下来，使用奇异值分解，得出逆换能器模型矩阵以及重建算法。使用所提出的算法重建声源和电导率图像。对理想点换能器和实际换能器进行了比较，以评估算法。最后，使用石墨体模进行了实验。我们发现，使用所提出的算法重建的声源图像比使用先前算法重建的图像匹配度更高，球形扫描几何结构的相关系数为98.49%，柱形扫描几何结构的相关系数为94.96%。理想点换能器和实际换能器的比较表明，球形扫描几何结构中的相关系数为90.2%，柱形扫描几何结构中的相关系数为86.35%。石墨体模实验的重建也表明，所提出的算法具有更高的分辨率。我们得出结论，所提出的考虑换能器特性的重建算法可以显著提高重建图像的分辨率。本研究可用于分析换能器位置和扫描几何结构对成像的影响。它可能为MAT-MI中电导率分布的重建提供一种更精确的方法。
