IEEE Trans Biomed Eng. 2018 Feb;65(2):282-293. doi: 10.1109/TBME.2017.2763590. Epub 2017 Oct 16.
Magnetic resonance electrical property tomography (MR-EPT) has significant potential for the estimation of the electrical properties (EPs) of tissue, which are essential for the calculation of specific absorption rates (SAR), a critical safety factor requiring monitoring and controlling in applications of ultrahigh field magnetic resonance imaging. In this paper, a novel, efficient method based on integral equations is proposed for the calculation of the EPs and the RF-coil-induced field. An inverse problem framework is first constructed to include the forward problem operator, while the EPs are reconstructed by using a nonlinear conjugate gradient method. The RF-coil-induced component is then calculated based on the achieved EPs and the forward operator. The proposed MR-EPT algorithm improves upon and differs from the existing methods in three aspects. First, a three-dimensional algorithm with improved efficiency is proposed. The higher efficiency arises from using a fast integral equation solver as well as an approximation of initial solution. Second, in addition to the EP values, the proposed method calculates the RF-coil-induced component, which is usually neglected in the existing MR-EPT algorithms. Here, we show that considering this field can significantly improve the accuracy of the SAR calculation. Finally, in contrast to differential approaches, the proposed method is more robust against noisy measurement of the transmit magnetic fields, because of the nature of the integral equations. The proposed method is verified through a full-wave simulation and an anatomically accurate numerical brain model, demonstrating its accuracy and efficiency.
磁共振电阻抗断层成像(MR-EPT)在估计组织的电学特性(EPs)方面具有重要的潜力,这对于计算特定吸收率(SAR)至关重要,SAR 是超高场磁共振成像应用中需要监测和控制的关键安全因素。在本文中,提出了一种基于积分方程的新型高效方法,用于计算 EP 和射频线圈感应场。首先构建了一个包含正向问题算子的逆问题框架,然后通过使用非线性共轭梯度法来重建 EP。然后根据获得的 EP 和正向算子计算射频线圈感应分量。所提出的 MR-EPT 算法在三个方面改进和区别于现有方法。首先,提出了一种效率更高的三维算法。更高的效率来自于使用快速积分方程求解器以及初始解的近似。其次,除了 EP 值外,所提出的方法还计算了射频线圈感应分量,这在现有的 MR-EPT 算法中通常被忽略。在这里,我们表明考虑该场可以显著提高 SAR 计算的准确性。最后,与差分方法相比,由于积分方程的性质,所提出的方法对发射磁场的噪声测量更具鲁棒性。通过全波仿真和解剖精确的数值脑模型验证了该方法的准确性和效率。