Abstracts

Rationale: Identification of the seizure onset zone (SOZ) based on visual inspection of the electrocorticographic (ECoG) recordings does not always result in a seizure-free outcome. Utility of high frequency oscillation (HFO) in identification of the SOZ has been shown to assist in this identification, in recent studies [1]. Although identification of the SOZ based on HFO has provided promising results, reliability and accuracy of this method has not been completely explored. The goal of this study was to develop an automatic method for identification of the SOZ based on HFO and investigate which frequency bands, from 70 Hz to 210 Hz, predict the surgical outcome more accurately.Methods: Three patients with epilepsy who underwent surgical treatment after evaluation with ECoG (intracranial video-EEG monitoring) were included. Resections were tailored individually based on visual inspection of the ictal patterns. Patients 1 and 2 were seizure free after surgery, but not Patient 3 (although this patient was improved). The ECoG data were recorded with the sampling rate of 1 KHz. For all patients, we identified the ictal onsets corresponding to all clinical seizures based on visual inspection of the ECoG data. We calculated the power of different frequency bands, from 70 Hz to 210 Hz, using the Hilbert transform (7 frequency bands). For each frequency band, we identified electrodes showing significant power enhancement during a 2 s time window immediately before the ictal onset. To this end, we first calculated the background cumulative distribution function (CDF) of power in each frequency band during 60 s time windows within five minutes before seizure onset. Then a value for power in each frequency band was calculated at 1 - 0.01 = 0.99 of CDF. After that, four fold power change was used as a threshold. Electrodes identified at that significant threshold were considered as SOZ electrodes recognized by HFO.Results: Our results show that the frequency band 150 - 170 Hz predicts the surgical outcome more accurately than other frequency bands (Table 1). In fact, complete resection of HFO electrodes resulted in seizure-free outcome in two patients in this frequency band. Moreover, the resected electrodes in Patient 3, who was not seizure-free, did not comprise some HFO electrodes in this frequency band. The high gamma frequency band, i.e. 70 - 90 Hz, has the smallest accuracy among all frequency bands in predicting the surgical outcome. In two seizure-free patients, HFO in frequency band 70 - 90 Hz identified 31 electrodes outside of the resection area, though these patients were seizure-free after resection.Conclusions: Our results show that HFO in the frequency band of 150 - 170 Hz predicts the surgical outcome. Identification of the SOZ based on HFO of the ECoG recordings may outperform the standard clinical procedure based on the visual inspection, and thus HFO should be included in decisions on surgical treatment. Acknowledgments This study was funded by the Children’s Foundation Research Institute & The Shainberg Neuroscience Fund, Memphis, TN. References 1. Epilepsia 2012; 53(9); 1607-1617.