Abstracts

Bicoherence quantitatively measures the degree of frequency coupling in a signal. Previous studies have shown that the electroencephalogram exhibits frequency coupling that varies with state of vigilance and with seizure onset. The hypothesis tested in this study is that the degree of frequency coupling increases as seizure onset is approached from the interictal-wake state. We studied 5 patients admitted consecutively to the Epilepsy Monitoring Unit for presurgical evaluation with intracranial depth electrodes for medically intractable temporal lobe epilepsy. This study was conducted with the approval and written consent of each patient. We collected sets of intracranial electroencephalogram in the interictal-wake state (defined as unequivocal wakefulness at least 2 hours before seizure onset or after seizure termination), the preictal-wake state (defined as unequivocal wakefulness immediately prior to seizure onset), the ictal-wake state (defined as a seizure occurring out of unequivocal wakefulness), and the interictal-sleep state (defined as slow wave sleep occurring at least 2 hours before seizure onset or after seizure termination). We performed a Kruskal-Wallis analysis to determine if the means of bicoherence summated over all frequency pairs (summated bicoherence) differed significantly between each of the above mentioned states. The summated bicoherence was significantly elevated in the preictal-wake state and the ictal-wake state with respect to the interictal-wake state in all 5 patients studied. In 3 of 5 patients, the degree of frequency coupling was significantly increased in the lead of seizure onset with respect to the other channels in the interictal-wake state. Summated bicoherence was also increased in the interictal-sleep state with respect to the interictal-wake state. Our results indicate that the degree of frequency coupling in the intracranial electroencephalogram, as measured by summated bicoherence, increases progressively from the interictal-wake state, to the preictal-wake state, and into the seizure state. These results suggest that summated bicoherence may be a useful metric in a seizure probability function. We also found increased frequency coupling in the seizure onset zone. Therefore, summated bicoherence may also prove to be a useful metric in identifying epileptogenic brain regions.