Abstracts

Rationale: Despite numerous studies conducted in human patients and animal models, the neurophysiological effects of deep brain stimulation (DBS) on cortical epileptic activity remain unclear [1]. Recently, in a patient suffering from partial epilepsy in relationship to a focal cortical dysplasia (CD) in the premotor cortex (PMC), we observed that DBS in the centromedian (CM) thalamic nucleus dramatically reduced the sustained paroxystic activity in the PMC [2]. Visual analysis of depth-EEG recordings showed that some stimulation parameters (frequency and amplitude) could induce a remarkable modulation of the dysplastic pattern whereas others did not affect the signals recorded in the PMC. This unique observation obtained in a human subject led us to quantitatively investigate the mechanisms involved in the modulation of epileptic activity by indirect thalamic stimulation. Our objective is twofold: i) to uncover the time-frequency content of depth-EEG signals recorded in the PMC before, during and after stimulation of the CM, and ii) to explain, from a neurophysiological viewpoint, this effects and their dependence on stimulation parameters, in particular the frequency.Methods: To quantify the time-frequency content of depth-EEG signals we used the Matching Pursuit method which operates signal decomposition into atoms (Gabor, Fourier, and Dirac). A feature vector representing the signal energy distribution in predefined frequency bands (delta to gamma) was then built using this decomposition. Feature vectors were computed on segments of depth-EEG activity. They were used to assess the similarities/differences among these segments, as recorded before, during and after stimulation and also for various stimulation parameters. Besides, we introduced a physiologically-relevant computational model of the cortex in order to identify some key parameter configurations (related to excitation, feedback and feedforward inhibition in the PMC) allowing for accurate reproduction of observed effects, in particular the suppression of the paroxystic activity in the CD.Results: Signal processing showed a reduction of at least 68% in the dysplastic slow pattern (2-4 Hz) for a stimulation at 2, 70, 100 and 150Hz but less than 8% at 50Hz. Moreover, an increase of energy in the beta_1 band (12-18Hz) was exclusively observed for high frequency stimulations. Computational modeling allowed us to generate hypotheses, at a neuronal population level, on the origin of these stimulation-dependant cortical modulations, as reflected by depth-EEG signals. Main findings include i) the specific behaviour of the principal cells in the CM nucleus in response to stimulation frequency and ii) the subtle role of collateral excitation and feedforward inhibition in the PMC in response to CM input.Conclusions: Stimulation of the CM can modulate electrical activity in a CD located in the PMC