Abstracts

Rationale: To determine if measures of phase synchronization may assist in identifying potentially epileptogenic cortical regions when applied to high density 256 channel interictal EEG recordings. Methods: Five patients with medically refractory epilepsy comprise the series. All were candidates for epilepsy surgery. In all cases definite localization of epileptic seizures was established on the basis of intracranial EEG recordings. Prior to invasive studies, all subjects underwent dense array 256 channel EEG recordings. Three minutes of interictal EEG were selected for analysis in all each case. The selected segments were at least two hours distant from an electrographic seizure and, based on visual analysis, free of interictal epileptiform patterns. Excessively noisy channels were removed and replaced with averages of surrounding electrodes. Data were imported into MATLAB for analysis. Phase synchronization was measured through repeated application of the Hilbert transform between pairs of EEG traces using a sliding window to encompass at least two cycles of the lowest frequency in the bandwidth examined (4-7 Hz, 7-12 Hz, 13-30 Hz, 30-50 Hz, 50-100 Hz). A synchronization index (SI) was established for each pair of electrodes, with a global index computed by comparison with nearby 6 electrodes. SI was calculated for a given channel based on these averages. Finally, long range temporal correlation (detrended fluctuation analysis) was calculated for the SI. The results were plotted on color intensity plots constructed from a 2-D layout of the electrode positions Results: In all five cases, electrode plots revealed regions of maximal phase synchronization that included regions where epileptic seizures were proven to originate. The most robust changes were observed in the 30-50 Hz range. The figure shows an example of individual with proven left lateral frontal lobe epilepsy. Note maximal synchronization in left lateral frontal lobe. Conclusions: Measures of phase synchronization in high density interictal EEG may be useful in identifying potentially epileptogenic regions.