Montage selection for temporal interference stimulation of the anterior cingulate cortex
DOI: Pending DOI
Synopsis
Transcranial temporal interference stimulation (tTIS) may enable noninvasive modulation of deep medial cortical targets, yet selecting a practical montage for bilateral anterior cingulate cortex (ACC) targeting remains challenging. We performed subject-level electric-field simulations across 93 two-pair tTIS montages (n = 9), ranking candidates with multi-objective exposure metrics that balanced ACC field strength, target enrichment, off-target gray-matter exposure, and superficial/deep-tissue penalties. Fz-T7 / FCz-P8 showed the most favorable balanced profile among feasible candidates, with stable ACC exposure in a validation cohort (n = 31). These findings support multi-objective bilateral ACC montage selection beyond target-intensity maximization.
Background
The anterior cingulate cortex (ACC) is a key medial cortical target involved in pain, affective regulation, cognitive control, and placebo-related modulation. Transcranial temporal interference stimulation (tTIS) has been proposed as a noninvasive approach for engaging deep or medial brain targets, but montage selection remains challenging because stronger target fields can be accompanied by broader off-target exposure 1–5. This challenge is particularly relevant for bilateral ACC targeting, where the goal is not simply to maximize ACC field strength, but to preserve bilateral ACC exposure while limiting high-field spread into non-target gray matter, deep white matter, and superficial frontal cortex. Existing montage workflows often emphasize unilateral or individually optimized target definitions, leaving open the question of whether a fixed two-pair montage can produce balanced bilateral ACC exposure across subjects. This study aimed to identify a practical fixed two-pair montage for bilateral ACC exposure using subject-level, multi-objective computational screening and independent validation.
Methods
We performed subject-level electric-field simulations using TI-Toolbox v2.2.4 with SimNIBS/CHARM head models generated from T1-weighted MRI scans 6,7. The discovery screen used 9 individual head models, and the selected montage was evaluated in an independent validation cohort of 31 additional individual head models, yielding 40 head models in total. Electrode centers were based on subject-specific SimNIBS 10–10 positions; candidate selection was subsequently constrained to locations available on the Neuroelectrics Starstim 10/10 64-position cap.
For each two-pair tTIS montage, the two carrier-pair fields were simulated as independent quasi-static transcranial direct current stimulation (tDCS)-like fields using circular 10-mm, 4-mm-thick electrodes and ±2 mA per pair (1:1). Vector fields were exported in native and Montreal Neurological Institute (MNI) space, and voxelwise temporal interference amplitude was summarized as maximal temporal-interference amplitude (TImax; V/m). Optimization focused on an a priori bilateral ACC region of interest (ROI) restricted to subject-specific MNI-space gray matter.
We used a two-stage montage-selection strategy. First, subject-specific flexible electrode searches were performed within a restricted frontocentral 10–10 electrode pool. Recurring high-performing electrode-location families were then converted into fixed, interpretable candidates and further constrained to available Starstim 10/10 cap locations for practical testing. For five high-performing montages, we additionally tested pair-1:pair-2 current-amplitude ratios of 1:1, 1:0.8, 0.8:1, and 1:0.6 while keeping electrode locations fixed. The discovery screen tested 93 montage conditions across 9 head models. The selected montage was assessed in a validation cohort (n=31), yielding a combined cohort of 40 head models.
Results
Montages were ranked using prespecified multi-objective TImax metrics computed from MNI-space maps. Primary target-exposure metrics were mean and 95th-percentile TImax (p95) within the ACC gray-matter (ACC-GM) ROI. Off-target gray-matter exposure and target-to-off-target specificity quantified the trade-off between ACC exposure and broader gray-matter spread. Top-field enrichment measured whether the highest-field gray-matter voxels were concentrated in the ACC-GM ROI, reported for the top 1% and top 5% of whole-GM TImax values; higher values indicate stronger ACC enrichment. Additional project-defined penalties quantified whether high exposure extended into deep white matter (deep-WM) or superficial frontal cortex. CANlab2024/Glasser parcels were used post hoc to profile regional exposure 8.
Across the discovery screen, montage ranking depended strongly on the selection criterion. Montages that maximized ACC-GM p95 were not necessarily optimal after accounting for target enrichment and off-target exposure. Posterior-temporal configurations produced stronger ACC exposure but showed broader field distributions and lower top-field enrichment, supporting multi-objective rather than ACC-GM p95-only selection. Figure 1A–C summarizes the selected Fz-T7 / FCz-P8 montage, group-mean (n=40) maximal temporal interference amplitude (TImax) map, and subject-level ACC-GM p95 distributions. The validation cohort showed a comparable ACC-GM p95 distribution to the discovery cohort, supporting exposure stability across individual anatomy.
Expanded screening identified Fz-FT9 / FCz-TP8 as the simulation-optimal balanced configuration, but this montage was not feasible on the target Starstim cap because FT9 was unavailable. We therefore evaluated Starstim-feasible 10–10 configurations for bilateral ACC exposure. Among these candidates, Fz-T7 / FCz-P8 (Figure 1D) provided the best overall balance of ACC target exposure, top-field enrichment, ROI/off-target gray-matter specificity, and superficial/deep-tissue penalties. Current-ratio steering in selected high-performing montage families did not improve the balanced exposure profile beyond the best fixed montage.
In the discovery screen, Fz-T7 / FCz-P8 ranked 4/93 overall and first among Starstim-feasible candidate montages by the prespecified composite score. It showed favorable target-exposure and specificity metrics, including median ACC-GM p95 of 0.660 V/m, top-5% enrichment of 15.50, top-1% enrichment of 24.87, and an ROI/off-target GM ratio of 1.346. Higher enrichment and ROI/off-target values indicate stronger ACC concentration and target-to-off-target specificity. Deep-WM and superficial-frontal penalties were relatively low at 1.203 and 0.944, where lower values indicate less off-target tissue exposure.
Bilateral parcel-level analysis showed highly balanced left–right exposure within CANlab/Glasser cingulate/ACC parcels, supporting Fz-T7 / FCz-P8 as the final balanced practical montage among completed tests without implying selective stimulation of ACC alone. Figure 1E shows the CANlab2024/Glasser parcel-level exposure profile of the selected montage. The highest-exposure parcels were predominantly located in bilateral ACC and medial cingulate territories, with the main non-target high-exposure parcels in right frontal opercular and dorsolateral prefrontal regions. Overall, the results indicate preferential enrichment of a bilateral medial cingulate system rather than isolated focal stimulation of a single ACC parcel. In the validation cohort, the selected montage maintained a comparable ACC-GM exposure profile, supporting anatomical generalizability beyond the discovery sample.
Discussion
These findings support a multi-objective framework for selecting tTIS montages for bilateral ACC targeting. Rather than maximizing target-field intensity alone, effective montage selection should balance target dose, bilateral exposure, top-field enrichment, and off-target penalties. The Fz-T7 / FCz-P8 montage provides a practical candidate for future experimental studies of pain, affective regulation, cognitive control, and placebo-related modulation. The results also suggest that simple current-ratio steering does not necessarily improve targeting when the goal is balanced bilateral medial cingulate exposure. The validation cohort supported generalizability across individual anatomy; future work should compare fixed versus individualized montage selection and test physiological and behavioral effects.
Disclosures
This project was supported by the National Institutes of Health (NIH) / National Institute of Mental Health (NIMH) under grant R37 MH076136. We acknowledge Amin Dehghani for contributing the 9 discovery head-model datasets and Marta Čeko for leading the larger dataset collection from which the validation cohort was derived. The authors report no commercial conflicts of interest related to this work.
References
1. Grossman N, Bono D, Dedic N, et al. Noninvasive deep brain stimulation via temporally interfering electric fields. Cell. 2017;169(6):1029-1041.e16. doi:10.1016/j.cell.2017.05.024.
2. Vassiliadis P, Beanato E, Wessel MJ, Hummel FC. Temporal interference stimulation for deep brain neuromodulation in humans. Nature Biomedical Engineering. 2026. doi:10.1038/s41551-026-01665-z.
3. Violante IR, Alania K, Cassara AM, et al. Non-invasive temporal interference electrical stimulation of the human hippocampus. Nature Neuroscience. 2023;26(11):1994-2004. doi:10.1038/s41593-023-01456-8.
4. Wessel MJ, Beanato E, Popa T, et al. Noninvasive theta-burst stimulation of the human striatum enhances striatal activity and motor skill learning. Nature Neuroscience. 2023;26:2005-2016. doi:10.1038/s41593-023-01457-7.
5. Dmochowski JP, Datta A, Bikson M, Su Y, Parra LC. Optimized multi-electrode stimulation increases focality and intensity at the target. Journal of Neural Engineering. 2011;8(4):046011. doi:10.1088/1741-2560/8/4/046011.
6. Haber I, Jackson A, Thielscher A, Hai A, Tononi G. TI-Toolbox: An open-source software for temporal interference stimulation research. Brain Stimulation. 2026;19(1):103016. doi:10.1016/j.brs.2025.103016.
7. Puonti O, Van Leemput K, Saturnino GB, Siebner HR, Madsen KH, Thielscher A. Accurate and robust whole-head segmentation from magnetic resonance images for individualized head modeling. NeuroImage. 2020;219:117044. doi:10.1016/j.neuroimage.2020.117044.
8. Glasser MF, Coalson TS, Robinson EC, et al. A multi-modal parcellation of human cerebral cortex. Nature. 2016;536(7615):171-178. doi:10.1038/nature18933.