Authors :
Presenting Author: Jay Jeschke, MA – New York University - Langone Health
Andrew Michalak, MD, MS – NYU Langone
Daniel Friedman, MD – Department of Neurology, New York University Grossman School of Medicine, NYU Langone Health
Elaine Sinclair, DO, PhD – VCU Health
Peter Rozman, MD – NYU Langone Health
Doris Xia, MD – NYU Langone Health
Patricia Dugan, MD – NYU Langone Health
Rationale:
Labeling and localizing stereoelectroencephalography (sEEG) contacts is a time-consuming process for clinicians, especially in cases involving complex or densely packed electrode arrays.
Methods:
We developed an automated pipeline to label SEEG electrode arrays using the post-operative CT, pre-operative T1 MRI, and the trajectory planning file from the stereotactic robot (ROSA, Zimmer-Biomet). The pipeline outputs an interactive HTML visualization that displays the patient's MRI with overlaid, color-coded electrode array labels. Accuracy was assessed in 10 consecutive SEEG patients. We then evaluated the time required for two trained epilepsy fellows to generate 3D CT images with labeled sEEG electrodes, with and without assistance from an automated labeling pipeline.
Results:
The algorithm produced accurate electrode array labels for 10 subjects (151 electrode arrays). The per-subject processing time ranged from 26-52 seconds. Clinical electrode array labeling times were significantly reduced with algorithm assistance (n=11, mean ± SD: 103 ± 62 seconds) compared to without assistance (n=12, 1222 ± 662 seconds), as confirmed by a Wilcoxon rank-sum test (p = 0.00024).
Conclusions:
Our simple-to-implement algorithm improves the speed of SEEG electrode array identification and labeling for manual electrode localization. The reduction in time to create labeled electrode images demonstrates substantial time savings and improved workflow efficiency.
Funding: None