Accuracy of frameless image-guided implantation of depth electrodes for intracranial epilepsy monitoring

Abstract

Various techniques are available for stereotactic implantation of depth electrodes for intracranial epilepsy monitoring. The goal of this study was to evaluate the accuracy and effectiveness of frameless MRI-guided depth electrode implantation. Using a frameless MRI-guided stereotactic approach (Stealth), depth electrodes were implanted in patients via burr holes or craniotomy, mostly into the medial temporal lobe. In all cases in which it was possible, postoperative MR images were coregistered to planning MR images containing the marked targets for quantitative analysis of intended versus actual location of each electrode tip. In the subset of MR images done with sufficient resolution, qualitative assessment of anatomical accuracy was performed. Finally, the effectiveness of implanted electrodes for identifying seizure onset was retrospectively examined. Sixty-eight patients underwent frameless implantation of 413 depth electrodes (96% to mesial temporal structures) via burr holes by one surgeon at 2 institutions. In 36 patients (203 electrodes) planning and postoperative MR images were available for quantitative analysis; an additional 8 procedures with 19 electrodes implanted via craniotomy for grid were also available for quantitative analysis. The median distance between intended target and actual tip location was 5.19 mm (mean 6.19 ± 4.13 mm, range < 2 mm-29.4 mm). Inaccuracy for transtemporal depths was greater along the electrode (i.e., deep), and posterior, whereas electrodes inserted via an occipital entry deviated radially. Failure to localize seizure onset did not result from implantation inaccuracy, although 2 of 62 patients (3.2%)-both with electrodes inserted occipitally-required reoperation. Complications were mostly transient, but resulted in long-term deficit in 2 of 68 patients (3%). Despite modest accuracy, frameless depth electrode implantation was sufficient for seizure localization in the medial temporal lobe when using the orthogonal approach, but may not be adequate for occipital trajectories.