Detecting Continuous High-frequency Oscillations in Intracranial EEG from Patients with Epilepsy

Epilepsy is a brain disease that manifests through recurrent seizures. In cases of refractory epilepsy, surgical removal of epileptogenic brain tissues can lead to seizure freedom. There are still 30% of patients who have seizures after surgery. High-frequency oscillations (HFOs) are regarded as a biomarker for localizing epileptogenic tissues, but HFOs are also found in normal brain regions. It is still challenging to distinguish pathological HFOs from normal HFOs in the clinical setting. We hypothesize that long-running rhythmic HFOs may indicate the pathology of brain tissue. This paper proposed a pipeline for detecting continuous HFOs (cHFO). First, we detect the HFO using the ensemble empirical mode decomposition (EEMD) method. Stereo-EEG from eight patients with epilepsy were analyzed, who achieved seizure-free outcomes after surgery. The HFOs with less than 100 ms intervals were further combined to form cHFOs. We analyzed the rhythmical index and amplitude fluctuation of those cHFOs. The minimum duration of cHFOs was chosen as 1 s, while those cHFOs can distinguish the pathological brain tissue best. Our results demonstrate that the proposed approach can detect cHFOs during preictal and ictal periods in epileptic tissue. Our study provides an automatic method for detecting cHFOs residing in pathological brain tissues, which may help improve the outcome of epilepsy surgery.