Abstracts

Rationale: A key challenge in delivering effective brain stimulation is choosing location and stimulation parameters that induce desired changes in neural activity, particularly when engaging complex neural networks. Our goal is to characterize the anatomical areas activated by stimulation of various sites in the temporal lobe, which can lead to guidelines for personalized stimulation targeting in the treatment of epilepsy and other neurological diseases.

Methods: Treatment-resistant epilepsy patients undergoing presurgical intracranial seizure monitoring were electrically stimulated at various locations, including the hippocampus, mesial and lateral temporal regions (n=54 patients, 158 stimulated locations), at amplitudes from 0.25-1mA and frequencies from 10-200 Hz. We designed a signal processing-based metric to quantify the strength of single-pulse evoked potential responses during the stimulation train, and used this metric to visualize whole-brain evoked responses as a function of stimulated location. Significant responses were defined by statistical comparison between stim-on vs. stim-off trials, and analyzed in terms of the anatomical locations of the recording electrode, stimulation site, and their proximity to white vs grey matter. 

Results: Significant evoked potentials demonstrated high signal-to-noise properties, commonly displaying small p-values (1e-20~1e-10) and consistent magnitude across trials within the same experiment. Responding locations were typically sparse and distributed (rather than appearing as clusters) in regions both near and far from the stimulated region. Stimulation in deep temporal regions (such as hippocampal, parahippocampal and fusiform areas) generally did not evoke responses in neighboring temporal contacts, and cortical stimulation responses were sparse and distributed widely over the cortical surface. Evoked potentials were also found to reliably predict post-stimulation changes in bandpower for high-frequency parameters, for the same stimulating and recording site. 

Conclusions: Anatomical connectivity appears to be a primary determinant of responses to electrical stimulation, even near the stimulation site. Evoked responses are predictive of functional changes in brain activity and are robust across trials for a given patient, stimulated contact, and montage, yet complex in their variability between patients, motivating the design of closed-loop strategies for personalized parameter optimization to treat epilepsy based on neurophysiological activation. 

Funding: NINDS U01NS113198