Local field potentials for target localization in centromedian deep brain stimulation for epilepsy

To evaluate whether local field potential (LFP) spectral profiles can serve as a candidate "spectral fingerprint" for physiological confirmation of centromedian-parafascicular (CM-Pf) targeting during thalamic deep brain stimulation (DBS) for drug resistant epilepsy (DRE).

This is a retrospective study of 10 patients (20 leads) who underwent CM-DBS implantation for DRE at a single tertiary center. Postoperative CT and preoperative MRI were co-registered, normalized to the Montreal Neurological Institute (MNI) space, and reconstructed using Lead-DBS software to anatomically localize contacts. BrainSense™ Survey recordings were obtained at least 3 weeks post-implant during routine programming. LFP frequency content was analyzed, and prominent peaks were identified and classified into canonical frequency bands (theta, alpha, beta). These spectral profiles were then mapped to MRI-based anatomical localizations, and statistical tests were applied to assess associations between peak patterns, contact localizations and thalamic subregions.

Contacts were distributed as follows: 50 in the CM, 16 in the Parafascicular (Pf), 15 in the Centrolateral, 8 in the Mediodorsal, and 7 in the Ventrolateral (VL) nuclei. Of the 10 representative spectral localizations confined to the CM/CM-Pf region, 8 (80%) displayed a distinct dual-peak spectral profile with peaks in the theta/low alpha (5.5-9 Hz) and high beta (20-30 Hz) bands (mean frequencies: 7.63 Hz and 21.02 Hz, Fisher's exact test, p < 0.001). Single-peak profiles showed no significant association with specific nuclei (p = 0.871). Contacts overlapping other thalamic nuclei more frequently exhibited narrow 10-15 Hz peaks (p = 0.005) or triple-peak profiles (p = 0.02), suggesting mixed structural contributions.

A dual- band candidate spectral pattern consisting of theta/low alpha and high beta peaks was associated with the CM-Pf region in this cohort. This finding provides early evidence supporting the feasibility of incorporating passive LFP recordings as a physiologic marker of target engagement. Future work to prospectively compare bipolar survey-based localization with monopolar recording strategies could enable development of a state-based, physiologically informed spectral atlas to refine CM-Pf targeting in thalamic neuromodulation for DRE.