Abstracts

Rationale: Our previous work has shown that by stimulating the brain with specific patterns of either sensory or direct electrical pulses, the so called carrier frequency paradigm, we are able to assess the ability of the system to generate epileptiform activity and eventually seizures. Computer model studies have shown that this effect can be explained on the basis of response properties of the neural network to stimulation and can account for multiple parameter-driven transitions to an epileptic seizure. We have reported on a, proof-of-principle, clinical application of this technique in patients undergoing pre-surgical evaluation with implanted electrodes both to localize the seizure onset site (SOS) and to indicate the time periods when seizures can be anticipated. The present contribution refines the carrier frequency modulation concept and excludes possible linear interference from volume conductance and/or stimulus artifacts while affording further statistical confirmation of our analysis technique based on the epileptogenic potential quantity called relative phase clustering index (rPCI). Methods: We applied a new, improved approach towards identifying the SOS in five patients and in an extra one to estimate the time of upcoming seizure. We used (1) bi-phasic cyclic alternating polarity stimulation sequences (BiCAP) and (2) bipolar EEG traces to compute rPCI . This improved approach excludes the influence of linear effects either from the stimulus or from the neural response. It also reduces possible influences from any common electrical reference. Results: We achieved correct SOS localization in all five cases, including an MRI-negative case in a patient with previous unsuccessful resection. The results of the long-term rPCI monitoring support the hypothesis that the seizures are associated with a bilateral increase of the rPCI. Conclusions: The study confirms our concept of fluctuation-driven transitions to epileptic seizures where only probabilistic predictions of the time of the seizure are possible. It also shows that non-linear neuronal dynamics account for the high rPCI values at the SOS. Our results warrant the use of rPCI as an identifier of seizure vulnerability and its possible use for state-dependent therapeutic approaches aiming at arresting epileptic seizures.