MRI metal artifact characterization in patients with spinal cord injury at 3 Tesla

(Phantom & Human) OBJECTIVE: The purpose of this study was to optimize imaging sequences to suppress artifacts induced by metallic hardware, using phantoms implanted with spinal hardware and participants with spinal cord injury (SCI) and spinal hardware at 3Tesla.

US METHODS: Magnetic resonance (MR) sequences were first tested on realistic agar-suspended spine phantom models with metallic instrumentation. C-spine with anterior-plate, posterior rods/screws, C-spine with rods/screws, C-spine with Kirschner wire, posterior T-spine with rods/screws, and anterior/posterior T-spine with anterior-plate and rods/screws were used. The optimized metal-suppression sequences obtained from phantom imaging were then evaluated on sixteen participants with SCI with similar metal implants. Four neuroradiologists performed a qualitative analysis and ranked all the scans, both with and without metal suppression. The following subjective visual assessment included: conspicuity of neural foramen, mitigation of artifact, visualization of the spinal cord and homogeneity of the cerebrospinal fluid (CSF).

Agreement between the raters was moderate (0.41 to 0.6) to substantial (0.61 to 0.8) for most measures, although some were in the fair range (0.21 to 0.4). In evaluating the T2 weighted-axial images for conspicuity of neural foramen, visualization of spinal cord, and homogeneity of CSF as well as T1 weighted-axial image for homogeneity of CSF in the anterior plate, the upper bound of the confidence interval was below "3" so the metal suppressed image was favored.

There is some improvement in using metal-suppressed sequences to evaluate spinal cord injury patients with metal hardware at 3T MRI; however, the model-adjusted mean scores did not reach statistical significance.