Abstracts

VERA: A User-Friendly Intracranial Electrode Localization Interface

Abstract number : 3.228
Submission category : 5. Neuro Imaging / 5A. Structural Imaging
Year : 2023
Submission ID : 925
Source : www.aesnet.org
Presentation date : 12/4/2023 12:00:00 AM
Published date :

Authors :
Presenting Author: Naasik Syed, BS – UT Southwestern

Markus Adamek, MSc – Washington University in St. Louis; James Swift, PhD – Washington University in St. Louis and National Center for Adaptive Neurotechnologies; Phillip Demarest, BS – Washington University in St. Louis; Peter Brunner, PhD – Washington University in St. Louis and National Center for Adaptive Neurotechnologies

Rationale:
Following intracranial electrode implantation for intractable epilepsy treatment, post-operative CT is coregistered to pre-operative MRI to identify electrode coordinates in MRI-space. Knowledge of electrode positions relative to anatomical structures is critical for both epileptogenic zone mapping and neuroscientific research. While there is no standard electrode localization method, current approaches to the problem have tedious workflows, require users to switch between multiple software platforms, and are monolithic or daunting to learn, creating a need for a tool with high usability and extendibility. Here we present an intuitive Versatile Electrode Localization Framework (VERA) integrating external tools into modular workflows that can easily be customized to meet users’ unique analysis specifications.



Methods:
VERA is implemented as a centralized MATLAB GUI through which users may interface with native and external tools, such as SPM12 or FreeSurfer. To test if VERA improves electrode localization accuracy, precision, and speed, 5 naïve users completed manual and VERA-assisted localization tasks, each composed of electrocorticography (ECoG) and stereo-EEG (SEEG) blocks. Participants completed 4 timed localization trials in random order to mitigate practice effects. Coordinates from an experienced user’s manual localization were used as ground truth electrode locations (n = 110 SEEG, 67 ECoG) and error magnitudes between user-placed points and ground-truth points were calculated. Left-tailed Wilcoxon rank sum tests were used to compare the median errors and durations of assisted and manual trial conditions. Levene’s test was used to compare the standard deviations (STDs) of error.



Results:
Using VERA improved SEEG electrode localization accuracy (median number of points ± STD = 107 ± 21 manual, 109 ± 3 assisted; median error = 0.48 mm manual, 0.41 mm assisted; Wilcoxon rank sum p < 0.01), precision (STD = 0.52 mm manual, 0.25 mm assisted; Levene’s p < 0.01), and speed (median time ± STD = 60 ± 21 min manual, 49 ± 20 min assisted, Wilcoxon rank sum p < 0.05). VERA also improved ECoG electrode localization accuracy (median number of points ± STD = 67 ± 4 manual, 67 ± 0 assisted; median error = 0.82 mm manual, 0.45 mm assisted; Wilcoxon rank sum p < 0.01), precision (STD = 1.17 mm manual, 0.45 mm assisted; Levene’s p < 0.01), and speed (median time ± STD = 59 ± 21 min manual, 11 ± 4 min assisted, Wilcoxon rank sum p < 0.05). These results demonstrate a strong improvement in electrode localization performance when assisted by VERA compared to manual localization.
Neuro Imaging