Abstracts

Rationale: We are interested in finding objective, quantitative differences between epileptiform spikes and benign variants. An eventual goal will be to understand and differentiate between the generators of epileptic versus benign cerebral activities. This study compares benign sharp spikes of sleep (BSSS) with spikes from temporal lobe epilepsy (TLE) patients. Methods: We chose three quantitative methods, principal component analysis (PCA) and single regional dipole source modeling (RSM), both provided by BESA, and distributed dipole modeling provided by LORETA. Thus far, we have considered six recent TLE patients with Engel Type I surgical outcome, and seven patients with BSSS, the latter with no other EEG abnormalities. For each patient we chose three clean representative spikes. In patients with bilateral spikes, we choose three from each side if possible. Highpass and notch filtering were applied judiciously to remove obvious low frequency activity and 60 Hz noise. Spatiotemporal PCA and RSM were applied to each spike over two different epochs centered around the first significant sharp peak of the spike. A short epoch was set to capture the peak alone, while the long epoch was set to capture the entire spike and wave complex. Since it generates temporally independent volumentric dipole results for each time point, LORETA was applied to the entire long epoch. Results: We considered the amplitude of the first five principal components for each spike, which generally accounted for over 99% of the variance. For BSSS, the first component accounted for an average of 88% of the variance for the short epoch, and 65% for the long epoch. For TLE, it accounted for 92% and 80%. Thus, for both short and long epochs, the first component explained less of the data for the BSSS vs. TLE spikes. Variance was shifted into the 2nd and 3rd components, which were larger for BSSS than TLE. RSM results showed regional generators in temporal lobes, sometimes deep and near the midline. Visually the BSSS sources were slightly deeper and closer to the midline than the TLE sources, but overlapped considerably. LORETA maps of distributed dipole activity have been produced for three each of the BSSS and TLE patients. Spatial concentrations of significantly active dipoles for BSSS were located more frontally on the underside of the temporal pole, while those for TLE were more posterior and lateral. Conclusions: PCA showed that the spikes of BSSS were more complex than TLE , needing more components to account for the spatiotemopral pattern. PCA may be robust enough to discriminate BSSS from TLE spikes. Neither RSM, nor distributed LORETA modeling were able to easily distinguish between the two, but it is possible that more sophisticated pre-processing and modeling would be able to. BESA and LORETA results were not inconsistent with each other, but were not as robust as the PCA results. Our results stand in contrast to a PCA of spikes in 39 children [Rodin 1995 Epilepsia 36: 1078-83], where the spikes associated with seizures were more complex than those in children with no clinical seizures.