Optimizing imaging parameters for MR evaluation of the spine with titanium pedicle screws

OBJECTIVE. This study examined the contribution and interdependence of multiple imaging parameters in clinical imaging sequences to aid practicing radiologists in minimizing artifacts during MR imaging of the spine after implantation of titanium pedicle screws. MATERIALS AND METHODS. A lumbar spine specimen with titanium pedicle screws implanted in the pedicle was imaged with a 1.5-T scanner. Sequence type, voxel volume, TE, and bandwidth varied. Different voxel volumes were achieved by altering section thickness, field of view (FOV), and matrix size. Artifact size was measured on sagittal and axial images at the midpedicle level. Artifact size was expressed as a percentage of actual screw size, and mean artifact size was calculated for each sequence. Analysis of variance without replication was done. RESULTS. Mean artifact size ranged from 231% to 364% of actual screw size. Artifact size was independent of voxel volume for voxels greater than 3 mm³ (p < .001). Artifact size decreased significantly [p < .001) when voxel volume was less than 1 mm³. When we increased slice thickness and maintained e constant voxel volume, artifact size decreased. Decreases in artifact size correlated with a reduction in the ratio of the FOV to the number of pixels in the frequency-encoding direction (N(x)). Artifact sizes were smallest when fast spin-echo sequences were used. Conventional spin-echo sequences produced artifacts that were smaller than the artifacts produced by the gradient-echo sequences. Decreasing the TE did not diminish artifact size for conventional spin-echo images at larger voxel volumes. CONCLUSION. Although voxel volume has been recognized as a factor that affects artifact size, the role of other contributing factors-slice thickness, number of phase-encoding steps, and FOV/N(x)-has not been evaluated before. Artifact reduction proved to be dependent only on FOV/N(x). Artifact size was reduced by the use of fast spin-echo sequences. With conventional spin-echo sequences, TE should be minimized, although other technical factors may outweigh the gain in artifact reduction.