Diffusion Tensor Imaging (DTI) in the axial view can be used for evaluating the function of the cervical spine with diffusion metrics and fiber tracking. Traditional single-shot EPI (ssEPI) is limited by low spatial resolution and geometric distortion. Multi-shot EPI (msEPI) with 1D-navigator and cardiac triggerring has been used for high resolution DTI in the cervical spine [1]. However, it is limited by residual artifacts due to incomplete phase correction and low acquisition efficiency due to triggering. In this study, 2D-navigated msEPI without cardiac triggering is used to achieve high resolution cervical spine DTI. A k-space reconstruction method, SYnergistic iMage reconstruction with PHase variatiOn and seNsitivitY (SYMPHONY), previously named SEPARATE [2], is used to correct the ghost artifacts caused by the motion-induced phase variations among different shots. In vivo data were acquired on 10 healthy volunteers to evaluate the image quality and quantitative measurement of the proposed technique.

Data acquisition

10 healthy volunteers (5 F and 5 M, age 23-28 years, mean 25.4 years) were recruited under IRB approval from our institution. All the data were acquired on a Philips 3.0T Achieva TX scanner (Philips, Best, The Netherlands) with a 16-channel SENSE neurovascular coil. For each volunteer, four sequences were conducted, including (1) T2-weighted (T2W) multi-echo FFE, (2) ssEPI DTI and (3) 2D-navigated interleaved EPI DTI (8 shots) with and (4) without peripheral pulse unit (PPU) triggering. b values of 0, 600 s/mm² and 15 diffusion directions were used in DTI scans. The detailed scan parameters were listed in Table 1. For all the scans, the same imaging locations covering C3-C7 were used. To be noted, the triggered multi-shot DTI was split into two successive stacks which covered the upper and lower cervical spine respectively, in order to include enough slices and to ensure TR similar with the non-triggered scan.

Image Reconstruction

Both triggered and non-triggered msEPI DTI data were reconstructed using SYMPHONY, which treats the phase variations among different shots as a power of encoding and recovers the k-space of each shot and channel in a GRAPPA way, with the interpolation weights calibrated from the 2D navigator.

Data analysis

The DTI parameters, including Fractional Anisotropy (FA), Axial Diffusivity (AD), Radial Diffusivity (RD) and Mean Diffusivity (MD), were calculated using DTIStudio [3]. After that, all the parameters were processed using ROI analysis in ROIEditor. The ROI drawing was demonstrated in Fig. 1. Firstly, one slice was chosen for each section of C3-C7. Secondly, the contour of whole spinal cord (Whole) was drawn on the mean DWI image to form the first ROI. Then the other three ROIs, lateral corticospinal tracts (CST), posterior columns (PC) and GM (grey matter) were drawn on the color-coded FA (cFA) map.

The triggered and non-triggered msEPI DWI images on one slice of one volunteer, with and without SYMPHONY reconstruction, are shown in Fig. 2. The ghost artifacts resulting from shot-to-shot phase variations are suppressed by SYMPHONY, for both triggered and non-triggered scan.

Fig. 3 shows the FA maps on different sections from one volunteer using ssEPI DTI, msEPI DTI with and without triggering. The T2W-FFE images are provided as anatomical references. In msEPI DTI, better delineation of white matter (WM) and GM can be obtained, due to the improved resolution; and the results are similar for the triggered and non-triggered scans.

Fig. 4 shows the averaged DTI parameters of all the volunteers, calculated from non-triggered msEPI for all ROIs in different sections. While the results in GM and WM ROIs are consistent with those obtained in previous studies using reduced FOV techniques [4, 5], AD and MD values in whole spinal cord turn out to be larger than those in both GM and WM. Further study is needed to investigate whether this is true or biased by CSF contamination.

In this study, high resolution cervical spine DTI is achieved using multi-shot EPI acquisition and SYMPHONY reconstruction. The proposed technique is validated using quantitative analysis on healthy volunteers. Because no cardiac triggering is used, the scan time can be reduced (by about 40% in this study). The improved resolution and scan efficiency are beneficial for the quantitative evaluation of the cervical spine in both neuroscience research and clinical diagnosis.