Abstracts

Rationale: Analysis of effective connectivity is poised to have wide spread implications in the future of the study of epilepsy, from epileptogenesis and ictal onset to seizure propagation and pre-surgical evaluation to outcome prediction. Various studies have compared different methods of connectivity in simulated data and in intracranial EEG in patients with intractable epilepsy, with findings suggesting that (i) the yield of the method of connectivity analysis may vary depending on the dimensionality and modal order of the data, (ii) yield of the method of connectivity analysis may also vary depending on whether the connections are linear or nonlinear in the epoch studied, and (iii) among multiple linear and nonlinear methods tested, no particular method outperformed the others among different models. Analyzing fluctuations in connectivity parameters in the various interictal and ictal states and seizure phases would help us to understand the complex and temporally dynamic interactions within seizure networks that evolve over the course of a seizure. In this study we evaluate the consistency of alteration in connectivity in semiologically similar seizures in patients with intractable focal epilepsy.Methods: The charts of all patients with intractable focal epilepsy who underwent an intracranial EEG study at the University of Pittsburgh were reviewed. All patients who underwent a surgical resection and who remained seizure free for one year post surgery were included in the study. Details regarding patient demographics, clinical, imaging and electrophysiological features were collected. Intracranial EEG epochs during wakefulness, stage 2 sleep, and the three most typical seizures semiologically and electrophysiologically were selected as visually identified by 2 epileptologists. The EEG was down sampled from 1024 Hz to 500 Hz and filtered to delta (1-4hz), theta (5-7hz), alpha (8-13 Hz), beta (15-30 Hz), gamma (30-100 Hz) and high frequency oscillations (100-250 Hz). Analyses was performed using MATLAB (v. 2012a) and HERMES. Effective connectivity was analyzed using correlation, coherence, imaginary coherence, directed transfer function, partial directed coherence and transfer entropy.Results: Results were quantitatively analyzed and compared. Significant differences were noted in partial directed coherence in various ictal and interictal states. Stage 2 sleep and awake states showed the least variation in connectivity as calculated by the various methods employed. The results are detailed in Figure 1.Conclusions: Significant differences were noted in connectivity in various ictal and interictal states, depending on the method of connectivity used, even with the same threshold parameter. A thorough understanding of the method used and the ictal and interictal phase analyzed is paramount in interpreting the results of connectivity analysis in patients with epilepsy.