Abstracts

Rationale: Similarities in the clinical manifestations of non-epileptic seizures (NES) and epileptic seizures (ES) can lead to erroneous diagnosis and treatment. Examination of EEG recordings from these patients can greatly enhance the likelihood of correctly identifying patients with ES. However, this diagnosis cannot be made conclusively without the presence of interictal or ictal epileptiform discharges, which may not be recorded during routine EEGs. As a result, multi-day EEG monitoring may be required to establish the diagnosis. Therefore, it would be beneficial to develop a diagnostic method that can reliably distinguish the EEG patterns of NES patients from those of patients with ES by analysis of brief interictal epochs of EEG. We investigated the existence of differences in signal characteristic dynamics of interictal EEGs recorded from NES and ES patients. Methods: Interictal EEG epochs (at least 10 seconds each) were sampled from long-term EMU recordings obtained from 14 patients - seven were diagnosed as ES patients with TLE and the remaining seven as NES patients. To reduce confounding effects due to variability during the interictal period, sampling of the interictal EEG epochs were constrained with the following three conditions: 1) no epileptic discharges; 2) no eye-blinking; and 3) clear alpha rhythmic activities in the occipital region. Therefore, all sampled EEG epochs were recorded during an eyes-closed and relaxed state without the presence of epileptic discharges. A total of 139 and 211 epochs were sampled from ES and NES patient groups, respectively. The mean signal regularity (using the pattern-match regularity statistic, PMRS) and inter-hemisphere signal correlation (using maximal cross-correlation, CCmax) over the temporal lobe region (i.e., F7, F8, T3, T4, T5, and T6) were calculated for each EEG epoch. Nested one-way ANOVA was applied to test the hypothesis that there exists a significant (p < 0.05) difference between the two patient groups with respect to these two signal characteristics of EEG dynamics. Results: The mean PMRS over ES patients was 0.461 (s.e. = 0.015) and was 0.496 (s.e. = 0.013) across NES patients (p-value = 0.018). The mean CCmax of ES patients was 0.283 (s.e. = 0.011) and was 0.333 (s.e. = 0.008) for NES patients (p-value = 0.007). Conclusions: These results suggest that EEG signals over temporal regions in ES patients are more regular but are less correlated between the two hemispheres. If these results hold in a large-sample study, it may be possible to develop a diagnostic tool that can enhance the usefulness of routine EEG recordings in the diagnosis of epilepsy.