Optimization of acquisition parameters of diffusion-tensor magnetic resonance imaging in the spinal cord

Objective: The purpose of this study was to optimize imaging parameters for diffusion tensor imaging (DTI) of the cervical spinal cord using a recently developed sensitivity-encoded (SENSE) imaging technique, which can substantially reduce susceptibility artifacts. Materials and Methods: One hundred twenty sets of DTIs were performed of the cervical spinal cord in 40 normal volunteers, using a SENSE-based echo-planar imaging technique with different parameters (b-values, numbers of diffusion gradient directions, number of excitations, and slice thickness) in a stepwise approach. In step 1, DTI was performed of the cervical spinal cord with different b-values 500, 700, 900 seconds/mm(2); then with different numbers of diffusion gradient directions 6, 15, 32 in step 2; different number of excitations 1, 3, 5 in step 3; and different slice thicknesses 2, 3, 4 mm in step 4. In each step, 30 sets of DTIs were obtained from 10 volunteers. To determine the optimal imaging parameters, 3 radiologists evaluated the qualities of fractional anisotropy (FA) maps and color FA maps by visual analysis. The number of reconstructed fibers was measured for quantitative analysis. All qualitative and quantitative comparisons were analyzed by statistical methods using the Friedmann test and the Wilcoxon signed rank test. Results: In step 1, DTIs using a b-value of 900 seconds/mm(2) showed the highest number of reconstructed fibers and the best image quality of FA map and color map. In step 2, the use of 15 or 32 directions demonstrated better quality DTIs than 6 directions. No significant difference was evident between the quality of DTI with 15 directions and that with 32 directions. The scan time of DTI with 15 directions was shorter than with 32 directions. In step 3, as the number of excitations increased, the number of reconstructed fibers increased significantly and the image quality of the FA map and the color map improved significantly. In step 4, the numbers of reconstructed fibers were significantly the highest with a slice thickness of 4mm. Conclusion: Optimal parameters for DTI in the cervical spinal cord included a b-value of 900 seconds/mm(2), 15 diffusion gradient directions, 5 excitations, and a slice thickness of 4mm.