MRI distortion correction produces geometrically relevant shifts in cisternal trigeminal nerve localization: implications for Gamma Knife radiosurgery

Abstract

Gamma Knife radiosurgery (GKRS) for trigeminal neuralgia (TN) requires millimeter-scale accuracy because a single 4-mm collimator targets the cisternal trigeminal nerve (CTN). MRI is essential for CTN visualization but is vulnerable to geometric distortion, particularly near the skull base. This study evaluated whether CT-based MRI distortion correction produces geometrically relevant changes in CTN localization. Methods Forty adult patients with normal intracranial anatomy who underwent cranial MRI and CT were retrospectively analyzed. Bilateral CTN segmentations were performed on uncorrected and distortion-corrected MRI by two independent raters. Segmentation masks were exported to 3D Slicer to extract center-of-mass (CoM) coordinates. The primary endpoint was patient-level 3D Euclidean CoM shift between corrected and uncorrected MRI. Dice Similarity Coefficient (DSC), axis-specific displacement, threshold positivity, inter-rater reliability, and Bland–Altman agreement were also assessed. Results The mean patient-level 3D shift was 1.12 ± 0.34 mm. All patients had at least one side-rater observation exceeding 0.5 mm, and 28 of 40 patients (70.0%) had at least one observation exceeding 1.0 mm. Mean DSC was 0.84 ± 0.03, indicating high but incomplete volumetric overlap. Displacement was anisotropic, with absolute magnitudes ordered as |ΔY| > |ΔX| > |ΔZ|. Inter-rater reliability was excellent, with ICC(A,1) values ranging from 0.927 to 0.988. Conclusion CT-based MRI distortion correction produces measurable, anatomically structured shifts in CTN localization. These shifts are non-negligible relative to the 4-mm GKRS collimator. Their clinical consequences remain unknown; dosimetric and outcome studies in treated cohorts are required to determine whether correction alters nerve dose coverage or clinical results.