S173. Automatic detection of high frequency oscillations (HFOs) to locate epileptogenic zone

Abstract

Introduction Surgical resection of epileptogenic zone is the first choice for the treatment of refractory epilepsy, and accurate localization of epileptogenic zone is the key to determine the outcome of the operation. In recent years, There are many research show that high-frequency oscillations (HFOs) is a promising biomarker for the epileptogenic zone. Visual HFOs evaluation is subjectivity and time-consuming. It seriously limits the application of the HFOs in clinic. Therefore, it is very important to develop high frequency oscillation automatic detection program to precisely locate the epileptogenic zone for improving the curative effect of epilepsy surgery. Methods We retrospectively studied EEG data from 20 patients at Xuanwu hospital, Capital Medical University in who underwent epilepsy surgery (from January 2017 to March 2017). Inclusion criterion: 1. Patients with complete medical records and underwent surgery resection. 2. Implanted subdural grids, depth electrode or stereotactic deep electrode (for SEEG). 3. Sampling rate > 2000 Hz. 4. For more than half a year’s follow-up. Exclusion criterion: 1. EEG data with obvious artifacts. 2. Patients with bilateral lesions. 3. Patients with occipital lobe epilepsy. Inclusion criterion of channels: HFOs rate > 1/min (the number of highest HFOs channels is less than 10/min, then all channels are analyzed). All patients signed informed consent and the project was approved by the Ethics Committee of Xuanwu hospital, Capital Medical University. We applied the automatic detection program based on maximum distribution peak point calculating the baseline which developed by our laboratory to perform HFOs analysis for all of patients. After that, we sort the channels from high to low according to the number of HFOs, then the resection ratio was calculated used the total number of channels which from the highest frequency channels to the first channel without resected divided the total number of produced HFOs channel. Finally, the resection rate was compared with the prognosis of the patients. Results Patients are in two groups based on Engel Classification. thirteen patients in good outcome which means Engel Ia (seizure free), and seven patients in poor outcome which means Engel Ib-IV (non-seizure free). In good outcome patients’ group, the resection ratio of ripples is 24% and fast ripple is 82% (the lowest ratio is 40%). In poor outcome patients’ group, the resection ratio of ripples is 6% and fast ripples is 12% (the highest ratio is 21.4%). We found that resection ratio of fast ripple has obvious statistical significance (P Conclusion Our results show that patients who remove higher ratio channels of the Fast ripple would have better surgical outcome, therefore, fast ripple is more accurate than ripple in localizing epileptogenic zone.