Thalamic stereoelectroencephalography: safety, accuracy, and thalamocortical connectivity

There is growing interest in thalamic sampling during epilepsy stereoelectroencephalography (sEEG). Data on thalamic sEEG safety, accuracy, and anterior (ANT), centromedian (CM), and pulvinar (PUL) nucleus connectomics is scarce. Here, we report the safety, accuracy and connectivity of thalamic sEEG and compare results to epilepsy DBS.

This single-center study included all patients with epilepsy implanted with thalamic sEEG or DBS between 2018 and 2024. Stereo-EEG surgery used a frameless stereotactic articulating arm, 3D printed stereotactic guide, or robot (sEEG lead diameter 0.8mm). DBS surgery used a rigid frame and cannula (1.81mm outer diameter). Post-surgical complications were identified by chart and imaging review. For each thalamic lead, accuracy was defined as the distance from the nearest electrode contact to the targeted nucleus Morel atlas volume (ANT, CM, or PUL), using an open-source toolbox. Estimated volumes of tissue activation/sampling were calculated for each most proximate contact for thalamocortical connectivity analysis. Connectivity analyses used a normative structural connectome, and connectivity patterns were compared.

160 thalamic sEEG leads and 188 DBS leads were implanted across 109 and 83 patients, respectively. One sEEG patient (0.9%) developed a symptomatic intraparenchymal hemorrhage with transient weakness of the contralateral upper extremity. Eight patients with DBS had transient post-surgical symptomatic complications, three with radiographic findings. Targeting accuracy was excellent for sEEG and DBS with median proximity of 0.30mm and 0.23mm, respectively; sEEG was associated with greater variability and more outliers (proximity≥3mm; 4% vs. 0, p=0.004). Thalamocortical connectivity patterns were highly consistent between sEEG and DBS cohorts with excellent overlap for ANT, CM and PUL subgroups (Spearman's ρ=0.86-0.98, p<1e^-5). The ANT subgroup showed preferential connectivity to prefrontal and mesial temporal cortices; CM with perirolandic cortex, supplementary motor area, and subcortical regions; and PUL with mesial and neocortical temporal, and parieto-occipital regions, with some territories of overlap or under-engagement.

Thalamic sEEG demonstrates a favorable safety profile, excellent targeting accuracy, and representative thalamocortical network engagement relative to DBS. Favorable sEEG safety profile may reflect difference in lead cross-sectional area (5.1-fold smaller than DBS cannula). Distinct ANT, CM, and PUL thalamocortical connectivity profiles support individualized, hypothesis-driven targeting.