Predictors of frame-based SEEG electrode implantation accuracy

Background

Factors affecting electrode accuracy in stereo-electroencephalography (SEEG) implantation have been inconsistently reported. In the setting of a recent expansion in stereotactic implantation techniques, we examined pre- and post-implantation variables influencing electrode accuracy, as well as factors contributing to unsuccessful sampling intended sub-lobar target from traditional frame-based method in order to provide a comprehensive reference for future implantation.

Methods

We analyzed consecutive patients who underwent frame-based SEEG implantations at two Australian centers. Pre-implantation MRI was co-registered with post-implantation CT to obtain planned and actual trajectories (PT/AT). Absolute target error (Euclidean distance), radial error (perpendicular to PT), and depth error (parallel to PT) were calculated between AT and PT. Pre- and post-implantation factors, including electrode trajectory and anatomical factors, were collected. AT were classified as off-target if they did not sample the intended cortical target/s. Multivariate generalized linear mixed model assessed factors associated with bone entry and radial errors. Binomial regression was employed to examine predictive factors for off-target electrodes.

Results

We collected data from 629 electrodes across 50 patients (12.58 ± 2.50 electrodes/patient). Median absolute, radial, depth errors, and bone entry point localization error (BEPLE) were 1.85 [IQR1.23–2.58], 1.56 [IQR0.95–2.26], 0.57 [IQR0.23–1.07] and 1.09[IQR0.74–1.45] mm, respectively. Fifty-four (8.59 %) electrodes were off-target and 19.89 %(41/207) of electrodes with radial error exceeded the 2 mm safety margin and were off-target. Of the pre-implantation factors, trajectory angle on the coronal plane (p = 0.01), bone thickness (p < 0.001), and implantation depth (p = 0.001) predicted radial errors. BEPLE, p < 0.001), bone thickness (p < 0.001) and implantation depth (p = 0.001) were significant post-implantation predictive factors. Off-target trajectories were associated with bone thickness (p < 0.001) and trajectory angle (p = 0.01) for pre-implantation and radial error (p < 0.001) for post-implantation variables.

Conclusions

This study analyzed predictive factors of electrode accuracy using a traditional frame-based technique to provide a comprehensive reference. Electrode target radial error is strongly predicted by increased implantation depth, bone thickness along the trajectory, and electrode angle at bone entry. Increased radial error is strongly associated with electrodes missing the intended sub-lobar target. Further research is required to investigate the full clinical implications of electrode inaccuracy.