Diffusion weighted imaging (DWI) is a proven marker of microscopic nervous system injury. However, imaging of small structures such as the spinal cord or peripheral nerve fibers remains challenging since sufficient resolution and signal to noise necessitates long scan durations¹. Likewise, quantifying diffusion parameters from structures of interest is routinely performed with manual region of interest analysis. To overcome these limitations, diffusion weighted spectroscopy was implemented in an attempt to monitor the tissue microstructure in a time-efficient manner. The use high diffusion weighting provided suppression of non-neural tissues. A spectroscopic localization and readout maintained sufficient signal to noise, albeit with the loss of spatial information. Several DWI schemes were compared, including conventional diffusion tensor², rotationally-invariant dPFG³, and DDE-filtering⁴ on the rat spinal cord in vivo.

Experiments were performed on 8-12 week old Sprague-Dawley rats using a Bruker 9.4 T Biospec System with a 4-channel surface array for reception and a 9 mm inner diameter volume coil for excitation. A Point Resolved Spectroscopy (PRESS) sequence was modified to include two pairs of Stejskal-Tanner diffusion sensitization gradients spanning each of the refocusing pulses. The diffusion gradients were independently adjustable in their duration (δ), separation (Δ), strength (b-value), and direction. For all PRESS experiments, a voxel (10x10x6 mm) was placed in the spinal cord aligned along its main axis with additional parameters: TE=41, TR=3s, number of points=256, spectral width=4960 Hz, outer volume suppression, and cardiac gating.

Initial DTI experiments for characterization of the diffusion-weighted behavior in the spinal cord utilized a 4-shot, respiratory-gated EPI sequence (TR/TE = 1500/28 ms) with 8 b-values and 2 directions oriented parallel and perpendicular to the cord axis. Subsequent DWI PRESS experiments used a diffusion duration and separation of 6 and 12 ms, respectively. Diffusion weighting directions were performed under 3 conditions: DTI: a single-axis DW scheme with 15 directions at a b-value of 2000 s/mm^2; dPFG: a double-axis DW scheme with 60 orthogonal and 12 parallel directions with b-values of 2000 s/mm² for each; DDE: a double-axis scheme with a constant perpendicular DW gradient at a b-value of 2000 s/mm² coupled with a DW gradient parallel to the cord with a range of 10 b-values between 0 and 2000 s/mm².

Each of the conditions were analyzed in Matlab with appropriate DW signal models: DTI: calculation of whole-voxel fractional anisotropy (FA); dPFG: calculation of the whole-voxel microscopic anisotropy (µFA); DDE: calculation of axial diffusivity (AD_filt) and the restricted fraction (f_R) fit with a biexponential model. For each spectra, signal was quantified by integrating the magnitude spectra from ± 800 Hz centered at water peak (no water suppression was used).

In the spinal cord, a diffusion weighted gradient applied perpendicular to the cord axis selectively attenuates the signal from all tissues except that of the cord white matter tracts presumably due to restricted diffusion in axons arranged perpendicular to the cord axis (Fig 1). Importantly, signal from other tissues such as muscle and cerebrospinal fluid (CSF) were attenuated to within the noise at b-values greater than 2000 s/mm². Thus, diffusion weighting of at least this value was employed for subsequent DW PRESS experiments. In a single animal, the values derived from the cord were DTI: FA=0.16, dPFG: µFA=0.33, DDE: AD=1.14, f_R=0.44. In DTI scheme, the largest eigenvalue was aligned along the cord (vector=[0.027 -0.047, 0.99]) demonstrating the DTI-type PRESS acquisition may be used to confirm voxel alignment with the cord. The orthogonally-invariant measurement (µFA), demonstrated a clear difference between the parallel and orthogonal measurements indicating high microscopic anisotropy.Measurements of diffusion in living nervous system tissues have the potential to be useful biomarkers of injury and disease activity. However, diffusion MRI has seen barriers to clinical integration due to the long scan durations and propensity for artifacts including motion. Structures such as the spinal cord and peripheral nerves have seen even more difficulty due to their limited size and complicating locations in the body that are prone to motion. This demonstration of feasibility of acquiring DW metrics using a PRESS localization is a step forward for more rapid measurements with the goal of advancing their clinical adoption.Alternatives to diffusion weighted imaging, including DTI, are needed to improve assessments in the spinal cord and peripheral nerves. The DWI schemes shown here have the potential to increase the efficiency of measurements, which may be important for clinical applications. While promising, the methods shown here require further validation in injury and disease models to demonstrate their ultimate application.