Abstracts

Rationale: Mounting evidence during the last few years proves that stereotactic EEG procedures are safer than subdural electrode implantations. Further, these approaches are more effective in localizing epilepsy, by a more comprehensive sampling of the epileptogenic network. The traditional French SEEG approach using mostly orthogonal placements does not always arrive at optimal coverage of the involved brain regions. The insula, ventral temporal cortex, temporal operculum and some of the medial fronto-parietal regions may not be optimally sampled and could be better studied using oblique trajectories, that entail steep angles of entry through the skull. neo-sEEG (nsEEG) is imaging based, robotically instantiated and open to the idea of sampling bias by MRI and prior recordings. Here, we performed an analysis of the relative safety and efficacy of oblique trajectories in nsEEG, comparing the accuracy of placement relative to shallower angle trajectories. Methods: A total of 2041 trajectories from 150 consecutive procedures performed by a single surgeon were analyzed. Oblique and azimuth based trajectories were predominantly used. Trajectories were planned using a T1 MRI with contrast and a CTA. Stereotactic depth electrode placement was performed using a stereotactic robot (ROSA; Zimmer Biomet, Warsaw IN) registered to the patient’s head using skull fiducials. Sub millimeter registration was obtained in each case. Preoperative and postoperative CTs were co-registered and deviations from the planned trajectories were used to quantify deviation of the entry and target points of each electrode. Angles of entry were quantified using custom Matlab scripts by creating a triangular face from the three closest vertices on the patient’s skull surface, and computing each trajectory’s deviation relative to a perpendicular at this face. A multiple linear regression model was then used to assess the impact of the registration error, chronological order of the procedure, electrode length, and angle of entry on the deviation of each electrode relative to the planned trajectory at the entry point and at target. Results: Mean root mean square (RMS) registration error for the patient population was 0.52 mm. Mean electrode working length was 46 mm. Mean and median of the entry site deviation were 1.32mm, 1.45mm and of the target side deviation were 1.47mm, 1.80mm. Deviations measured at both the entry and target points were significantly but minimally influenced by working length (β=-0.0065, p<10-4; β=0.013, p<10-6) and to an even smaller extent by angle of entry (β=0.0046, p=0.01; β=0.0086, p=0.005). Thus both the working length and the angle of entry had minimal impact on the deviation of each electrode from its planned trajectory. 109 of the total 150 patients underwent a resection following their implantation and 61% of the patients that had follow up data reported an Engel Class I outcome at last follow up (median duration since surgery = 25 months). Conclusions: SEEG accuracy is only minimally affected by electrode angle of entry and working length. Stated differently, oblique trajectories are only marginally less accurate than orthogonal trajectories The advent of 3D planning and visualization platforms for SEEG allows us to be creative in the design of trajectories to target putative seizure networks. While such designs need to be directed by clear hypotheses, it does not appear that we need to be constrained to classic orthogonal or azimuth based trajectories, and with appropriate surgical technique, oblique trajectories can be safely used. Funding: No funding