Pathologies of the spinal cord, such as those due to trauma, neurodegenerative diseases, and cancers, are a primary cause for functional disability (paralysis) and chronic pain. Though MRI already plays a role in the evaluation of spinal cord pathology, it continues to be hampered by image artifacts owing to magnetic field inhomogeneity, underscored in a recent review as “the greatest challenge for acquiring MR images in the spinal cord.”¹ Underlying the field inhomogeneity are a number of distinct local susceptibility inclusions such as cartilage, vertebral bones, fat, and the oxygen of the nearby airways and lungs, the latter of which acquires a time-varying component by virtue of the subject’s respiration. This study reports the first results applying a specially designed 24-channel shim array to image the human spinal cord.Hardware: The shim system consists of 24 independently driven, rectangular-planar electrical coils, atop which sits the custom-built 8-channel transceiver array. These two coil arrays, first described in an earlier work,² insert into the patient bed-table (Fig. 1) so as to lie in close proximity to the subject’s spine. Acquisition: For this initial proof-of-concept experiment, two healthy subjects (S1 & S2) were scanned on a 3T system (Magnetom Tim Trio, Siemens Healthcare, Erlangen, Germany). Standard 2nd-order shimming was first performed using the Siemens shims. High resolution T2-weighted anatomical images were obtained of the spinal cord with parameters: TE/TR=4.92/30 ms (S1), 3.88/15 ms (S2); flip angle 10°; isotropic voxel size 1.1 mm³; FOV=55x220x220 mm³ (S1), 46x280x280 mm³ (S2). For field mapping, a multi-echo gradient-echo sequence was acquired with parameters: TE=[4.92, 7.38, 11.13, 14.88] ms, TR=200 ms; flip angle 50°; spatial resolution=2.2x2.2 mm² in-plane, with 8 sagittal slices of thickness 3.0 mm, for an effective FOV=24x176x282 mm³. To assess the static field correction specifically, Subject 2 was asked to inspire and maintain apnea during field map acquisition (approx. 10 s). Processing: To create binary masks for 3d phase unwrapping,³ the corresponding magnitude images were set to zero for intensities below 1 percent of the maximum. Unwrapped phase differences between the first two echoes were normalized by the echo time difference and scaled to Hz to obtain the field maps. The high resolution anatomical images were passed to the sct_propseg tool of the Spinal Cord Toolbox (http://www.neuro.polymtl.ca/downloads) to delineate the shim volumes of interest (VOI) automatically in about 30s.⁴ Spinal cord VOIs as well as the shim reference maps² were interpolated to the grid spacing of the field maps prior to shim optimization. Shim optimization was rapidly performed using the fmincon function of the MATLAB optimization toolbox, minimizing the sum of the shim fields and the measured field map over the VOI while accounting for non-linear system constraints (e.g. max current per channel, and max total current).⁵ Optimal shim currents were set and the field maps were reacquired, during which Subject 2 was asked to once again maintain an inspired apnea.Results are shown in Fig. 2. The segment of the spinal cord targeted for shimming was approximately 7 cm³ and 15 cm³ in subjects 1 and 2 respectively. Shimming over these volumes using the the 24-channel array reduced the standard deviation of the field from 49.19 Hz to 22.85 Hz for Subject 1 (an improvement of 53.6 %), and from 13.65 Hz to 9.36 Hz for Subject 2 (an improvement of 31.4 %).The greater degree of field inhomogeneity observed for Subject 1 was likely due in part to less effective initial shimming using the standard shims in combination with (and perhaps more importantly) the presence of respiration-induced field variation. This respiration-induced variation was reported to be on the order of 100 Hz in the spinal cord across levels T1-T8 at 3.0 T.⁶ Work is underway to synchronize shim updates to the subject respiration to correct for these distortions in real-time. Though the results remain preliminary and limited to two subjects, the very-high order shim insert offers a promising technique to improve EPI-based quantitative MRI and spectroscopy of the spinal cord.