MRI-Only Preoperative Localization and Semiautomated Three-Dimensional Modeling of the Cervical Vagus Nerve for Vagus Nerve Stimulation Implantation.

BACKGROUND

We evaluated whether three-dimensional steady-state free precession with time-spatial labeling inversion pulse (3D-SSFP with time-SLIP) MRI could localize the cervical vagus nerve before vagus nerve stimulation (VNS) implantation and developed training-data-free semiautomated vagus nerve tracking software.

METHODS

Thirty patients with drug-resistant epilepsy underwent preoperative cervical 3D-SSFP with time-SLIP MRI followed by VNS implantation. The left vagus nerve was visually identified at the planned surgical level, and its relationship to the common carotid artery (CCA) and internal jugular vein (IJV) was classified. MRI localization was validated against intraoperative findings. A MATLAB-based tool semi-automatically tracked the vagus nerve, CCA, and IJV from user-defined reference points. Accuracy was assessed using coordinate error and classification agreement.

RESULTS

The vagus nerve was identified on MRI in all 30 patients, with complete concordance between MRI and intraoperative localization. Among 29 classifiable patients, the nerve was located superficially between the CCA and IJV in 45%, superficially and lateral to the CCA in 41%, deep between the CCA and IJV in 10%, and deep and medial to the CCA in 3.4%. The median distance between software-derived and visually identified coordinates was 1 mm (interquartile range, 1-2 mm), and software-derived classification agreed with consensus visual classification in 27 of 29 patients (93%).

CONCLUSIONS

Cervical MRI enabled preoperative localization of the vagus nerve and adjacent vessels. The semiautomated method showed close agreement with expert visual assessment and may support anatomical orientation and surgical planning for VNS implantation.