0063
Multi-Dimensional Diffusion MRI of the Human Sciatic Nerve1Institute of Imaging Science, Vanderbilt University, Nashville, TN, United States, 2Radiology, Harvard Medical School, Boston, MA, United States, 3Radiology, Brigham and Women's Hospital, Boston, MA, United States, 4Random Walk Imaging AB, Lund, Sweden, 5Philips Healthcare, Markham, ON, Canada, 6Philips Healthcare, Dallas, TX, United States, 7Radiology and Radiological Sciences, Vanderbilt University Medical Center, Nashville, TN, United States
Synopsis
Multi-dimensional diffusion MRI is a promising new tool to assess tissue microstructure. Here, we developed a high-resolution protocol that overcomes the challenges of nerve imaging in humans (e.g., the influence of fat, short-T2s) and translated the method to the sciatic nerve in the thigh. Preliminary results showed that measures of microscopic fractional anisotropy in sciatic nerve were 1) similar to reported measures in white matter and 2) were repeatable across scans in the same subject. Additional comparisons to conventional diffusion encodings of the sciatic nerve suggest that multi-dimensional diffusion MRI may yields more specific measures of nerve microstructure.
Introduction
DTI in peripheral nerves [1] based on conventional diffusion encodings cannot distinguish inflammation, edema, de/remyelination, Wallerian degeneration, and partial volume effects—all of which can occur concurrently and lead to similar changes in fractional anisotropy (FA). This impedes the ability to relate these measures to pathological changes in the peripheral nervous system. Multi-dimensional diffusion encoding employs diffusion encoding along multiple spatial directions for each shot [2,3,4], disentangling the effects of orientation and restricted diffusion through measures such as the microscopic fractional anisotropy (µFA). This additional information may improve discrimination between myelin and axonal pathologies, the latter of which is predictive of long-term outcomes, during nerve degeneration and regeneration. Multi-dimensional diffusion has not been applied in the peripheral nervous system due to the technical challenges associated with nerve imaging, including the competing need for resolution and SNR, the influence of fat around the nerves, and the short-T2 of the nerves. In this study, we 1) developed new multi-dimensional diffusion waveforms and acquisition protocols to overcome these obstacles, 2) tested the protocol in the sciatic nerve via scan-rescan studies, and 3) compared results to a conventional DTI protocol.Methods
Subjects: Six healthy volunteers were scanned lying feet-first and supine, with imaging performed in the thigh of the right leg, above the knee. Three subjects returned for a repeat scan approximately one week after the first scan. Data acquisition: Subjects were imaged with a 3.0-T Philips Achieva dStream MR scanner using a 16-channel knee T/R coil. The multi-dimensional diffusion protocol for spherical (STE) and linear tensor encoding (LTE) used the optimized waveforms [5], each of which were compensated for concomitant fields [6] and designed to minimize TE (Figure 1). The maximum b-value = 1300 s/mm2 used herein was lower than in previous white matter protocols due to the larger diameter of axons in nerves, allowing for further reductions in TE to compensate for the short-T2s of nerve and the need for higher SNR in our high-resolution nerve protocols. Double fat suppression with SPAIR and gradient reversal was applied to minimize the effect of fat around the nerves. Additional parameters included: b-values = {0, 100, 500, 900, 1300} s/mm2, 45 diffusion-encoding directions, TR/TE = 4500/85 ms, resolution = 1.5x1.5x10 mm3, averages = 1, slices = 8, scan time = 10 minutes total for both STE and LTE scans. The conventional DTI protocol included: max b-value = 900 s/mm2, 16 diffusion-encoding directions, TR/TE = 4000/62 ms, resolution = 1.5x1.5x9.6 mm3, averages = 6, slices = 8, and scan time = 10 minutes. Data analysis: All data were co-registered via 2D affine transformations using FSL. STE and LTE data were processed using diffusional variance decomposition [4,7] in order to estimate µFA. Conventional DTI data were processed via Camino to estimate FA. Regions of interest (ROIs) were then manually selected on all slices for the sciatic nerve and mean±SD slice-wise values were estimated for µFA and FA.Results
As shown in Figure 2, significantly lower group measures were observed for conventional FA (FA±SD=0.54±0.03) relative to µFA from the multidimensional diffusion protocol (µFA±SD=0.95±0.08). Additionally, a paired-sample t-test between measures of FA and µFA for all subjects was found to be significant (p<0.001). For the scan-rescan subjects, group measures for µFA were similar between the first scan (µFA±SD=0.95±0.07) and second scan (µFA±SD=0.97±0.03). These differences were quantified via the coefficient of variation (COV = ratio of the SD to the mean), which is was 3.4% for µFA across all slices. Representative slices of thigh and sciatic nerve are shown for T1-weighted anatomical scans, conventional FA maps, and µFA maps (Figure 3). Note that µFA was homogenous within the sciatic nerve (red box), while estimates of muscle µFA were less stable. More detailed plot of powder averaged mean ROI signal intensities and model fits are provided in Figure 4. Note the separation between the STE/LTE curves in nerve, which is indicative of a higher µFA. In contrast, very little separation between STE/LTE curves was observed for muscle.Discussion & Conclusion
We have demonstrated that, despite the challenges of nerve imaging in humans, multidimensional diffusion in the sciatic nerve is feasible and provides repeatable across scans. Future research will focus on better fat suppression techniques (e.g., methods that account for olefinic fat resonances), optimization of the b-value/direction sampling schemes, optimization of the registration and post-processing pipelines, and applying this method in a larger cohort of controls and patients with peripheral neuropathies. Future studies will also investigate the relative contributions of modeling and microstructural tissue parameters to the observed differences in FA and µFA in peripheral nerve.Acknowledgements
R01 NS097821 for funding.References
1. Noguerol TM, Barousse R, Socolovsky M, Luna A. Quantitative magnetic resonance (MR) neurography for evaluation of peripheral nerves and plexus injuries. Quantitative Imaging in Medicine and Surgery. 2017;7(4):398-421.
2. Shemesh N, Jespersen SN, Alexander DC, Cohen Y, Drobnjak I, Dyrby TB, et al. Conventions and nomenclature for double diffusion encoding NMR and MRI. Magnetic Resonance in Medicine. 2016;75:82-87.
3. Cory DG, Garroway AN, Miller JB. Applications of Spin Transport as a Probe of Local Geometry. Abstracts of Papers of the American Chemical Society. 1990;199:105.
4. Lasič S, Szczepankiewicz F, Eriksson S, Nilsson M, Topgaard D. Microanisotropy imaging: quantification of microscopic diffusion anisotropy and orientational order parameter by diffusion MRI with magic-angle spinning of the q-vector. Frontiers in Physics. 2014;2:11.
5. Sjölund J, Szczepankiewicz F, Nilsson M, Topgaard D, Westin CF, Knutsson H. Constrained optimization of gradient waveforms for generalized diffusion encoding. Journal of Magnetic Resonance. 2015;261:157-68.
6. Szczepankiewicz F, Nilsson M, editors. Maxwell-compensated waveform design for asymmetric diffusion encoding. ISMRM 2018; Paris, France.
7. Szczepankiewicz F, van Westen D, Englund E, Westin CF, Stahlberg F, Latt J, et al. The link between diffusion MRI and tumor heterogeneity: Mapping cell eccentricity and density by diffusional variance decomposition (DIVIDE). Neuroimage. 2016;142:522-32.
Figures
