Abstracts

Rationale: This study uses independent component analysis (ICA) to investigate the dynamic interactions of independent dipoles in status epilepticus. Using ICA and dipole fitting algorithm, the mixed dynamic signals of scalp recorded electroencephalography (EEG) could be statistically separated into independent dipoles, which are presumed to be the cerebral generators of the recorded EEG. These independent dipoles could transiently show coherence, such as during an epileptic discharge. The complex dynamics between the independent dipoles in status epilepticus could be characterized to provide a new measure of quantitative EEG analysis of status epilepticus. Methods: Retrospective EEG recordings from patients (n=6) with generalized status epilepticus were selected according to the approved protocol by the institutional review boards. These EEGs were processed in EEGLAB (http://www.sccn.ucsd.edu/eeglab/). After computing ICA, the components of interest in status epilepticus were identified. New EEGs were reconstructed and compared to the original EEGs to visually review the adequacy of component selection. Dipole fitting of these components was performed and mapped on a 2-d or 3-d head model. The dynamic interactions of these components were reviewed on a video clip. Correlation analysis between any two selected components was computed at selected time periods. The correlation coefficients were averaged across subjects, and were compared with data with similar analysis from patients (n=11) with interictal epileptic spikes for control. Results: The EEG in generalized status epilepticus showed increased coherence between the identified dipoles. In addition, the increase of coherence between dipoles was observed throughout the status epilepticus. This is in marked contrast to the increase of coherence that was noted in interictal spikes, in which it was only seen during the spike discharges, but not in between spikes. The means of the correlation coefficients of the none-spike periods in the control, the spike periods in the control and the status epilepticus were 0.003 (n=11), 0.115 (n=11) and 0.097 (n=6) respectively. The average of the s.d. across subjects for these coefficients were 0.001 (n=11), 0.109 (n=11) and 0.084 (n=6). Conclusions: ICA analysis showed a significant increase of coherence during generalized status epilepticus between multiple independent generators of EEG in the brain. The increase of coherence shown by ICA might provide a quantitative measure to represent hyperexcitability in EEGs from status epilepticus.