Abstracts

Rationale: With improved computing power and advanced algorithm development there has been a push to analyze EEG data in a way that previously was not possible by human review of EEG recordings. The goal of these studies are to both uncover basic mechanisms of seizure generation and network dynamics as well as find patterns that can be used to improve seizure freedom and decrease morbidity after epilepsy surgery. There has been tremendous excitement to study the discrete temporal and spatial occurrence of high frequency oscillations, but there is also burgeoning work on a number of approaches to analyze the continuous data stream of the EEG. In the past, epileptologist have used spikes to localize abnormal tissue, but the data is lacking for using spikes to localize seizure onset or understand the underlying network dynamics. With this gap in knowledge, we are using computer algorithms to determine if spike timing, location, and temporal patterns can improve seizure onset localization and ultimately outcome in patients undergoing epilepsy surgery. Methods: This study is IRB approved. The entirety of intracranial EEG from 7 pediatric patients (from 48-330 hours) who underwent phase 2 surgical evaluations was analyzed for patterns of spike occurrence, propagation, and location. These patterns were compared to a host of clinical variables and to other quantitative measures of the EEG. Clinical variables used were seizure type, seizure onset pattern, seizure onset zone, Engel class outcome, tissue resected, and pathology. Linear and non linear regressive models as well as non parametric statistical measures (bootstrapping) were employed to determine if patterns of spiking existed and the relationship of these patterns to the clinical variables. Results: Spike frequency both increased and decreased as the delta frequency content of the EEG shifted. No clear correlation to percent of delta or alpha activity was present. Surprisingly, spike frequency was found to modulate in a very slow oscillation (millihertz) in all patients. In addition, we found that spikes propagated in discreet patterns across grids and these patterns did not occur randomly but were recorded reproducibly over the duration of the recording. Conclusions: Interictal spike activity has not been fully evaluated for its ability to improve seizure onset localization and for understanding local network dynamics that may drive seizures. Our preliminary results suggest that delta activity can modulate spike frequency and spikes oscillate over time. Local network patterns may be observable by studying local spike dynamics. Further work on these patterns may shed light onto what tissue to resect to improve outcome in children undergoing epilepsy surgery.