Reproducibility of the MRI-defined spinal cord position in stereotactic radiotherapy for spinal oligometastases

Abstract Purpose To establish the reproducibility of the MRI-defined spinal cord position within the spinal canal. Materials and methods We acquired T1- and T2-weighted MRI scans of 15 volunteers on spine levels C7, T8 or L2. The scan protocol was repeated several times for different postures and time intervals. We determined the spinal cord shift (LR, AP, CC) using a rigid, grey value, vertebral body registration, followed by a spinal cord registration. We tested the sensitivity of our method, introducing artificial spinal cord shifts by varying the size and direction of the water-fat-shift (WFS) of the MR sequences. Results The spinal cord position on MRI is reproducible within approximately 0.2mm SD (LR, AP) and 0.7mm SD (CC) when reproducing the posture on the same day, as well as several weeks later. However, when comparing different postures, shifts of ∼1.5mm were found. Varying the WFS difference between scans (0.6–3.0mm) induced equivalent virtual spinal cord shifts (0.5–2.5mm). Conclusions Displacement of the spinal cord inside the spinal canal may occur as a result of posture change. Considering the total geometric accuracy of spine SBRT, MRI-defined spinal cord position is sufficiently reproducible and requires no addition to the typical setup-and-intrafraction motion PRV margin if posture is identical throughout the RT process.