Abstracts

Rationale: Intracranial recordings in patients with epilepsy provide a means to study normal and abnormal hippocampal neurophysiology. Given the deep subcortical location and unique anatomy of this structure, in vivo recordings are usually accomplished with penetrating linear (one-dimensional) depth electrodes. This renders inconsistent and sparse spatial sampling of potentially important neurophysiological features. We assessed the feasibility of novel microelectromechanical system fabrication approaches using highly conformal, thin-film micro-grid arrays to sample homogenously over the two-dimensional hippocampal surface, intra-operatively in patients undergoing epilepsy surgery. Methods: Five patients with medically-refractory seizures (n=5) and one with a lateral temporal lobe glioma (n=1) underwent limited excision of the antero-lateral temporal lobe for clinical purposes, which involved opening the lateral ventricle and exposing the hippocampal surface. A 32-channel micro-grid array (4 x 8 layout, 2.0mm spacing, 1.22mm diameter platinum/iridium contacts; Lawrence Livermore National Laboratories) was carefully placed and conformed on the dorsolateral hippocampal surface (facing the temporal horn). Hippocampal activity was recorded at 24 kHz from all 32 contacts for approximately 5-10 minutes, under general anesthesia in four patients (Right-sided) or an awake state in 2 patients (Left-sided). Results: Signal correlation between electrodes of the power in theta (4-7) and high-gamma (70-150) bands decreased with increasing distance and was fit to a Lorentzian model (Theta R2 0.348+-0.157; High Gamma R2 0.519+-0.150). In two of the six subjects, the angle between electrodes provided unique information compared to distance alone when predicting the signal correlation of the power in theta (p=0.001, p=0.02, based on 10,000 shuffles) and one of the six subjects in high gamma (p=0.006; based on 10,000 shuffles). Epileptiform discharges and pathological high-frequency oscillations were observed in all patients, except the patient undergoing resection for a lateral temporal tumor. The two awake patients also performed a visual naming task, producing increased gamma (30-70 Hz) and theta power from 100-300ms after picture display, and for approximately 200ms before speech onset (stronger activity in the former). During the naming task gamma activation traveled posterior-to-anterior, and theta oscillations traveled in a posterior-medial to antero-lateral direction (Rayleigh circular average of inter-electrode coherence; p<0.001), with negligible change in angle between stimulus (encoding) and pre-speech (knowledge retrieval) periods. Conclusions: Micro-grid recordings enabled a two-dimensional spatial perspective of neurophysiological activity over the human hippocampal surface, which provided novel information regarding signal correlation and spatiotemporal activation patterns. This not only confirmed that gamma activation and theta waves travel along the hippocampal axis, but that their direction (angle) of travel of the latter is oblique to the posterior-anterior direction concluded in recent studies. This technology has significant long-term potential for investigating the normal and pathophysiological patterns of activity in the human hippocampus. Funding: Dr. Kleen was funded by NINDS (R25NS070680-07), and Dr. Chang by the NIH (R01-DC012379, R00-NS065120), DARPA (DP2-OD00862) and the Esther A. and Joseph Klingenstein Fund.