Accuracy of Depth Electrodes is Not Time-Dependent in Robot-Assisted Stereoelectroencephalography in a Pediatric Population

Abstract

BACKGROUND AND OBJECTIVES: Robot-assisted stereoelectroencephalography (sEEG) is steadily supplanting traditional frameless and frame-based modalities for minimally invasive depth electrode placement in epilepsy workup. Accuracy rates similar to gold-standard frame-based techniques have been achieved, with improved operative efficiency. Limitations in cranial fixation and placement of trajectories in pediatric patients are believed to contribute to a time-dependent accumulation of stereotactic error. Thus, we aim to study the impact of time as a marker of cumulative stereotactic error during robotic sEEG. METHODS: All patients between October 2018 and June 2022 who underwent robotic sEEG were included. Radial errors at entry and target points as well as depth and Euclidean distance errors were collected for each electrode, excluding those with errors over 10 mm. Target point errors were standardized by planned trajectory length. ANOVA and error rates over time were analyzed using GraphPad Prism 9. RESULTS: Forty-four patients met inclusion criteria for a total of 539 trajectories. Number of electrodes placed ranged from 6 to 22. Average root mean squared error was 0.45 ± 0.12 mm. Average entry, target, depth, and Euclidean distance errors were 1.12 ± 0.41 mm, 1.46 ± 0.44 mm, −1.06 ± 1.43 mm, and 3.01 ± 0.71 mm, respectively. There was no significant increased error with each sequential electrode placed (entry error P-value = .54, target error P-value = .13, depth error P-value = .22, Euclidean distance P-value = .27). CONCLUSION: No decremental accuracy over time was observed. This may be secondary to our workflow which prioritizes oblique and longer trajectories first and then into less error-prone trajectories. Further study on the effect of level of training may reveal a novel difference in error rates.