Synopsis

Keywords: Data Acquisition, Data Acquisition

To overcome acquisition downfalls of HERMES edited magnetic resonance spectroscopy at 3T, we have conducted critical optimization for four clinically relevant brain regions in neurological disorders by finding the optimal voxel rotation that affects gradient order to enhance data quality and reproducibility.

Introduction

Have you ever had to discard data in your study where participants’ recruitment is limited? In our large post-traumatic stress disorder study, we had to discard over fifty percent of edited MRS data due to the poor SNR (<3), spurious echos, and dephasing of unwanted coherences. These artifacts could arise from insufficient, anisotropic B0 field distribution, suboptimal slice selection localization, and of course water and air in sinuses and mouth. To reduce unwanted artifacts, robust volume localization is necessary for both single voxel edited and unedited MRS ³. Here, we are focusing on data acquisition guidelines for edited MRS, specifically HERMES at Siemens Prisma 3T in hippocampus, insula, anterior cingulate cortex, dorsolateral prefrontal cortex, and insula. Identifying optimal voxel position and angulation that affects gradient order is crucial for acquiring high quality data and minimizing out of volume spurious signals as seen in Figure 1.

Methods

The study was approved by the Institutional Review Boards (IRBs) of Purdue University and informed consent was obtained. Twenty participants (16 males, 4 females, age: 33.7±8.3) underwent an hour-long brain scan with a whole-body 3T Siemens PRISMA MRI system (Siemens Healthineers, Erlangen, Germany). A vendor-supplied 64-channel receiver head coil was used across all participants. Four brain regions: anterior cingulate cortex (ACC), dorsolateral prefrontal cortex (DLPFC), insula, and hippocampus (hippo) have been investigated as a part of the larger study assessing alterations in neurometabolic profile of post-traumatic stress disorder. 3D T1-weighted images (MPRAGE, TR/TE = 2300/2.98 ms, slice thickness = 1.2 mm, field-of-view = 256 × 256 mm²) were acquired to select a volume-of-interest (VOI). Edited MRS was acquired using Hadamard Encoding and Reconstruction of MEGA-Edited Spectroscopy (HERMES)² with TE/TR:80/2000ms, 256 avgs, and VOI:40x25x27mm³. Data was analyzed and visualized for spectral evaluation in Gannet. Systematic variations in VOI rotation that affect gradient order for localization were evaluated for optimal acquisition. Data was discarded if it did not meet the standards of consensus papers as well as visual inspection where frequency drift, spurious echos, lipid contamination, and motion were observed in fitted spectra.

Results

Optimal slice position, voxel rotation that resulted in different slice selection gradient order was determined for four brain regions of interest. Figure 3 illustrates where parameters can be found on Siemens Prisma 3T scanner, while Figure 4 illustrates our recommended findings with included image reference for all four brain regions.
Slice selection order could be determined from combination of Sagittal (S), Coronal (C), and Transverse (T) order.
For ACC we found the most consistent results in S-T-C order and a VOI rotation >50º.
In DLPFC the T-C-S order or S-T-C order are recommended.
In hippocampus S-T-C order with VOI rotation >60º or T-S-C is optimal.

Lastly, in insula S-T-C or S-C-T order at positive angles is recommended.

Discussion and Conclusion

Optimal voxel rotation and slice selection order allowed us to significantly decrease the amount of data we discarded due to spurious echoes or artifacts as shown in Figure 2. Our future studies will include simulating gradients and cross-checking our findings with other edited sequences across GE and Siemens scanners.

Acknowledgements

We would like to thank Purdue University Northwest, Weldon School of Biomedical Engineering, Leslie A. Geddes Graduate Fellowship, and Bottorff Fellowship for their funding and support.