Advanced Imaging Research Center, University of Texas Southwestern Medical Center, Dallas, TX, USA.
Russell H. Morgan Department of Radiology and Radiological Science, The Johns Hopkins University School of Medicine, Baltimore, MD, USA; F. M. Kirby Research Center for Functional Brain Imaging, Kennedy Krieger Institute, Baltimore, MD, USA.
Neuroimage. 2019 Nov 15;202:116141. doi: 10.1016/j.neuroimage.2019.116141. Epub 2019 Aug 31.
Edited magnetic resonance spectroscopic imaging (MRSI) is capable of mapping the distribution of low concentration metabolites such as gamma-aminobutyric acid (GABA) or and glutathione (GSH), but is prone to subtraction artifacts due to head motion or other instabilities. In this study, a retrospective motion compensation algorithm for edited MRSI is proposed. The algorithm identifies movement-affected signals by comparing residual water and lipid peaks between different transients recorded at the same point in k-space, and either phase corrects, replaces or removes affected spectra prior to spatial Fourier transformation. The method was tested on macromolecule-unsuppressed GABA-edited spin-echo MR spectroscopic imaging data acquired from 8 healthy adults scanned at 3T. Relative to non-motion compensated data sets, the motion compensated data had significantly less subtraction artifacts across subjects. The residual choline (Cho) peak in the spectrum (which is well resolved from as a different chemical shift from GABA and is completely absent in a spectrum without subtraction artifact) was used as a metric of motion artifact severity. The normalized Cho area was 5.14 times lower with motion compensation than without motion compensation. A 'removal-only' version of the technique is also shown to be promising in removing motion-corrupted artifacts in a GSH-edited MRSI acquisition acquired in 1 healthy subject. This study introduces a motion compensation technique and demonstrates that retrospective compensation in k-space is possible and significantly reduces the amount of subtraction artifacts in the resulting edited spectra.
编辑后的磁共振波谱成像（MRSI）能够绘制低浓度代谢物（如γ-氨基丁酸（GABA）或谷胱甘肽（GSH））的分布图谱，但由于头部运动或其他不稳定性，容易出现减法伪影。在这项研究中，提出了一种用于编辑 MRSI 的回顾性运动补偿算法。该算法通过比较在 k 空间相同点记录的不同瞬变之间的残留水和脂质峰来识别受运动影响的信号，并在进行空间傅里叶变换之前，对受影响的光谱进行相位校正、替换或去除。该方法在 3T 扫描的 8 名健康成年人的未受大分子抑制的 GABA 编辑自旋回波磁共振波谱成像数据上进行了测试。与非运动补偿数据集相比，运动补偿数据集在受试者间的减法伪影明显减少。谱中的残留胆碱（Cho）峰（其与 GABA 的化学位移不同，并且在没有减法伪影的谱中完全不存在）被用作运动伪影严重程度的指标。与没有运动补偿相比，归一化 Cho 区域降低了 5.14 倍。还显示了该技术的“仅去除”版本在去除 1 名健康受试者的 GSH 编辑 MRSI 采集中的运动相关伪影方面也很有前景。本研究介绍了一种运动补偿技术，并证明了在 k 空间中进行回顾性补偿是可行的，并显著减少了编辑谱中减法伪影的数量。
