Abstract
Short-lasting rhythmic activity in intracranial electroencephalogram (iEEG) in the frequency range of 80 to 500 Hz is regarded to be a promising biomarker of epileptogenicity. This activity is referred to as high-frequency oscillation (HFO), and its detection from iEEG is considered the first step to several applications. In this study, several multi-channel algorithms for HFO detection are proposed. With the proposed multi-channel statistics and threshold determination scheme, the algorithms allow HFO detection to be performed without breaking the iEEG channel structure and the detection threshold to be determined automatically. Experimental simulation results illustrate the advantage of the proposed algorithms over existing single-channel-based approaches.