Comparison of multicenter MRI protocols for visualizing the spinal cord gray matter

Abstract

Purpose: Spinal cord gray-matter imaging is valuable for a number of applications, but remains challenging. The purpose of this work was to compare various MRI protocols at 1.5 T, 3 T, and 7 T for visualizing the gray matter. Methods:In vivo data of the cervical spinal cord were collected from nine different imaging centers. Data processing consisted of automatically segmenting the spinal cord and its gray matter and co-registering back-to-back scans. We computed the SNR using two methods (SNR_single using a single scan and SNR_diff using the difference between back-to-back scans) and the white/gray matter contrast-to-noise ratio per unit time. Synthetic phantom data were generated to evaluate the metrics performance. Experienced radiologists qualitatively scored the images. We ran the same processing on an open-access multicenter data set of the spinal cord MRI (N = 267 participants).Results: Qualitative assessments indicated comparable image quality for 3T and 7T scans. Spatial resolution was higher at higher field strength, and image quality at 1.5 T was found to be moderate to low. The proposed quantitative metrics were found to be robust to underlying changes to the SNR and contrast; however, the SNR_single method lacked accuracy when there were excessive partial-volume effects. Conclusion:We propose quality assessment criteria and metrics for gray-matter visualization and apply them to different protocols. The proposed criteria and metrics, the analyzed protocols, and our open-source code can serve as a benchmark for future optimization of spinal cord gray-matter imaging protocols.

Optimized multi-echo gradient-echo magnetic resonance imaging for gray and white matter segmentation in the lumbosacral cord at 3 T

“…In our analyses, echoes acquired at echo times above this range (at 19 and 23 ms) did not improve GM segmentability further. Notably, these findings are in line with the results obtained in simulated cervical cord images, where CNR between GM/WM reached a plateau at echo times between 15 and 17 ms 36 . In any case, the shortest possible echo time, echo spacing, and repetition time should be selected to accommodate the maximum number of echoes and signal averages within the available imaging time.…”

“…GM and WM have different T1 (GM: 994 ms, WM dorsal column: 853 ms, WM lateral column: 830 ms) and relative proton density values (GM: 0.8, WM: 0.7), yielding a good contrast already in the first echo. However the contrast increases only slightly with increasing echo times due to similar T2* values at 3 T (GM: 41.3 ms, WM: 39.8 ms) 17 , 35 , which is not enough to compensate for the steady loss of SNR at later echoes (also demonstrated by Cohen-Adad et al 36 ). This overall leads to a steady decreased in CNR with increasing echo time (also shown in Barry and Smith 17 ).…”

Body size interacts with the structure of the central nervous system: A multi-center in vivo neuroimaging study

“…101,102 Different associations of CSA-GM and CSA-WM with other investigated variables, may affirm the necessity of further development of MRI protocols imaging SC GM in high contrast and detail. 103…”

“…101,102 Different associations of CSA-GM and CSA-WM with other investigated variables, may affirm the necessity of further development of MRI protocols imaging SC GM in high contrast and detail. 103 Recently, the correlation between CSA-SC at C2-3 level and body height, body weight, brain (WM/GM) volumes and thalamus volume were observed in 804 UK Biobank brain imaging database participants. 21 Our current spine-generic database study complements the UK Biobank results and expands the knowledge that these observations are almost exclusively SC WM-related.…”

Advanced spinal cord MRI in multiple sclerosis: Current techniques and future directions

“…SC GM area is best measured on T 1 -weighted phase-sensitive inversion recovery (PSIR; Papinutto et al, 2015 , Schlaeger et al, 2015 , Schlaeger et al, 2014 ) or gradient-echo T 2 *-weighted axial images, both of which provide good GM/WM contrast. Recent work comparing different sequences and field strengths for GM visualization ( Cohen-Adad et al, 2022 , Papinutto and Henry, 2019 ), as well as a framework for the quantitative evaluation thereof, paves the way for further development of better SC GM structural imaging.…”