Spinal cord MRI techniques are emerging that can characterize aspects of microstructure: diffusion tensor imaging (DTI), magnetization transfer (MT), and T2*-weighted imaging (T2*-WI). Degenerative cervical myelopathy (DCM) is a common condition involving multilevel disc degeneration, often with hypertrophy and/or ossification of ligaments within the spinal canal, leading to cord compression (Figure 1). In this prospective longitudinal study, we develop a clinically feasible multimodal MRI protocol, establish normative data, and investigate if these techniques can quantify injury to individual tracts rostral to cord compression, correlate with global and focal disability, and predict outcomes in DCM.37 DCM patients (age 56.2; 54% male; 19 mild, 12 moderate, 6 severe; 2 with previous surgery) were enrolled consecutively from outpatient neurosurgery clinic. 29 age/gender-matched healthy subjects served as imaging controls. Data collection included detailed clinical assessments (mJOA, Nurick, MDI, QuickDASH, Berg Balance (BB), grip, upper extremity (UE) strength, modified GRASSP, GaitRITE), electrophysiology (ulnar SSEP), and quantitative MRI (3T GE) with 13 axial slices covering C1-C7 (Figure 2): single-shot EPI DTI (TR/TE=4050/75ms, b=800s/mm2, 25 directions, 1.25x1.25x5mm voxels, 3 acquisitions averaged offline, 2m06s each), axial 2D SPGR (TR/TE=36/5.7, FA 6, NEX=3, 1x1x5mm voxels)+/- MT pre-pulse (3m43s each), and MERGE T2*-WI (TR/TE=550/5,10,15, 0.6x0.6x4mm voxels, 4m15s), co-registered to a T2-weighted image (T2-WI) (6m30s) (Figure 3). Data collection will be repeated at 1-year follow-up. Semi-automated analysis was performed using the Spinal Cord Toolbox¹ for segmentation, motion correction, registration to WM atlas, and extraction of metrics with automatic correction for partial volume effects (Figure 1). Reliability was analyzed by calculation of intra-class correlation coefficient (ICC) for 7 MT datasets (2 measurements), 66 DTI datasets (3 measurements), 13 DTI datasets using a 3-run average (2 measurements), and 8 T2*-WI datasets (2 measurements). 5 a priori metrics were selected: cord cross sectional area (CSA) at C1, fractional anisotropy (FA), mean diffusivity (MD), MT ratio (MTR), and T2*-WI WM/GM ratio (a novel metric representing grey-white contrast). Regions of interest (ROIs) in the dorsal columns (DCs) and lateral corticospinal tracts (LCSTs) at C1-C3 were also selected a priori, to avoid potential bias at compressed levels. Univariate/multivariate analysis of variance (ANOVA/MANOVA) was performed to compare DCM patients vs. controls. Spearman correlations were calculated between metrics and measures of global disability (mJOA, BB) and focal impairment (mJOA motor scores, UE strength, grip force, sensation).Reliability of quantitative assessments was excellent for all 4 metrics extracted from DCs and LCSTs (ICCs: 0.86-0.99) (Table 1). Metrics from individual DTI scans showed lower reliability (0.82-0.94), but averaging over 3 acquisitions improved reliability substantially (0.93-0.99). DCM patients demonstrated evidence of WM injury rostral to the compressed spinal cord (multivariate p<0.0001), including decreases in CSA, FA (DCs, LCSTs), and MTR (DCs, trend in LCSTs), and increased T2*-WM/GM ratio (DCs). MD did not show differences vs. controls. Cross-correlations of MTR, FA, and MD metrics were moderate, ranging from 0.2 to 0.5, whereas T2*-WM/GM ratio did not correlate closely with other metrics. FA provided the most robust results overall, showing highly significant differences vs. controls (p<0.0001 in DCs, LCs) and the strongest correlations among MRI metrics with all clinical measures (Table 2). The strongest correlations were also frequently found in the expected anatomical tract, with FA of the bilateral LCSTs correlating well with mJOA motor scores was (r=0.64), FA of the ipsilateral LCST predicting arm power (left: r=0.44, right: r=0.53) and grip strength (left: r=0.57, right: r=0.54), FA of the DCs correlating well with ipsilateral sensation (left: r=0.66, right: r=0.67), and FA in the DCs correlating with BB (r=0.51) (all p<0.05). Furthermore, 2 patients who had undergone previous surgery with metallic implants (lower cervical spine) showed data of acceptable quality at C1-C3.We have developed a reliable multimodal spinal cord MRI protocol that is feasible to implement in a clinical environment, using standard clinical hardware and with an acquisition time of approximately 30 minutes. Bias has been minimized by utilizing automated tools for quantitative assessment and focusing on the rostral portion of the cervical cord, avoiding difficulties with cord segmentation and susceptibility artefact in the compressed cord and enabling assessment of patients with surgically implanted hardware. WM injury of individual tracts can be successfully quantified in the rostral spinal cord, correlating well with focal neurological deficits and global impairment. Each of our imaging techniques appears to provide complimentary information, and multivariate linear regression analysis of a larger cohort is underway.Our results demonstrate that accurate and reliable quantitative assessment of white matter damage can be achieved with clinically relevant methods, which could be of benefit for all neurological disorders affecting the spinal cord.