Human intracranial signal stability tracks anatomical accuracy after electrode reimplantation

Leriche, Ryan B.1 2 ; Wozny, Thomas A.2 3; Saal, Jeremy2; Motzkin, Julian C.2; Sellers, Kristin K.2 ; Starr, Philip A.2; Chang, Edward F.2 ; Shirvalkar, Prasad 2

Published Jun 16, 2026 on Dryad. https://doi.org/10.5061/dryad.8931zcs3r

Data files

Jun 16, 2026 version files 8.05 MB

README.md

4.51 KB
staged_iEEG_stability_repo.zip

8.05 MB

Abstract

Deep brain stimulation (DBS) for neuropsychiatric disorders increasingly relies on patient personalized approaches. One strategy to identify viable targets for personalized stimulation and biomarker detection uses a temporary, inpatient stereoelectroencephalography (sEEG) trial to identify optimal targets for permanent re-implantation. However, this approach reasons that neural signals are stable between repeated intracranial electrode implants, which is not yet validated. We characterize the spectral stability of local field potentials after electrode reimplantation from an ongoing single-center clinical trial using two-staged sEEG/DBS for chronic pain (NCT04144972). Repeat neurosurgical implant of intracranial electrodes reliably targeted intended trajectories across staged surgeries occurring months apart (2.1 ± 1.2 mm Euclidean distance across 17 brain targets, 5 participants). Anatomical targeting error was correlated with differences in recorded electrophysiological power spectra between stages (r(15) = 0.51, p = 0.03). This correlation was stronger for the subset of regions re-targeted exclusively for biomarker sensing (r(6) = 0.73, p = 0.04), and non-significant for therapeutic stimulation sites (r(7) = .23, p = 0.76). Similarly, a multiple regression model significantly predicted spectral distance (F(4, 13) = 5.99, p = 0.008, adjusted r = 0.70) with an interaction between electrode type and Euclidean distance (t (13) = 2.39, p = 0.03). Chronic intracranial EEG signals recapitulated temporary recordings in 17 of 17 brain regions with the major spectral peak within 1 Hz. Our results demonstrate the temporal stability of neural signals across months and validate the use of a two-staged intracranial EEG approach for chronic DBS sensing in humans.

Dataset DOI: 10.5061/dryad.8931zcs3r

Description of the data and file structure

Local field potentials were recorded from inpatient invasive EEG (iEEG) and ambulatory iEEG.

These are technically known as stereoelectroencephalography and deep brain stimulation electrodes, respectively.

5 participants with chronic pain had a range of 20–159 inpatient iEEG recordings, and 13–99 ambulatory iEEG recordings.

Files and variables (within staged_iEEG_stability_repo.zip)

Within ~/staged_iEEG_stability_repo/data data for each Figure is included in .xlsx files with the prefix figX_XXX, except Figure 1A which was created in Adobe Illustrator.

Throughout ambulatory and inpatient are abbreviated as ambul and inpat corresponding to the stage where the data were collected.

Files and their sheets

Each file is the top-level bullet point while sheets within that file are sub-bullet points.

fig1_elec_coord.xlsx
- inpat_elec_coord:each row contains the 3D coordinates of a given electrode in subject native space in mm (columns — "Slicer_x, Slicer_y, Slicer_z"). The column "Subject" refers to which specific subject the electrode pertains. The columns "Location", and "G_W" provide greater detail and what region the brain the electrode was and whether it was labelled grey vs white matter, respectively.
- ambul_elec_coord: same column naming and data as inpat_elec_coord, but for ambulatory stage.
- staged_iEEG_euclid_distance:the nearest inpatient and ambulatory stage contact pairs per region. Note how participants have 2–4 regions. The column "Subject " refers to the specific subject while "region" refers to the unified brain regions naming scheme used between inpat and ambul stages. The columns "inpat_channel" and "ambul_channel" are categorical data referring to the electrode pair which are nearest each other while the column "min_euclidean_distance" specifies their minimum euclidean distance in mm.
fig2_P1_spec_cohort_spec_euclid_corr_data.xlsx
- inpat_spec_freqs:the frequency bins corresponding to inpatient spectra in Hz
- inpat_P1_YYY_spec:inpatient neural spectra corresponding to subject P1. YYY corresponds to the brain region shorthand and it's specific bipolar contacts. Rows are individual sessions, columns are frequencies from inpat_spec_freqs,and cells are the periodic power in log10(Power) units.
- ambul_spec_freqs:same column naming and data asinpat_spec_freqs, but for ambulatory stage.
- ambul_P1_YYY_spec same column naming and data as inpat_P1_YYY_spec, but for ambulatory stage.
- cohort_spec_euclid_dist similar to staged_iEEG_euclid_distance, but with simplified regions and spectral distance such that the Euclidean distance to spectral distance can be explicitly interrogated. The column "spectral_distance" is the absolute difference between the mean inpat and mean ambul spectral power per region in log10(Power). The column "inpat_lead_role" is categorical data of whether a specific region was targeted for neural sensing or therapeutic stimulation—which is used in a multiple regression later on. For completeness, the column "days_btwn_implant" is the number of days between inpatient iEEG explant and ambulatory iEEG implant per region.
fig3_P1_center_freq_cohort.xlsx
- inpat_P1_YYY_center_freq the center frequencies of the major peak per session (i.e., the frequency where the highest amplitude spectral peak occurs). YYY is the region shorthand.
- ambul_PX_YYY_center_freq same but for the ambulatory stage
Sfig2_P2_P3_spec_center_freq_data.xlsx and Sfig3_P4_P5_spec_center_freq_data.xlsx
- Same naming schemes and as in Figures 2 & 3, but with more participants/regions using "PX" to refer to different participants.

Code/software

The following toolboxes were used on MATLAB version 2022a to run this repository:

Statistics and Machine Learning Toolbox     Version 12.3        (R2022a)

Start from MAIN_wrapper.m, all functions and paths are automatically added/explained within:

~/staged_iEEG_stability_repo/code

Functions plot per figure sub-panel under the figX_XXX naming convention and are called from MAIN_wrapper.m.

Human subjects data

I confirm that subjects consented to publishing de-identified data. We removed subject names and utilized numbered subject IDs.