Automated detection and classification of high frequency oscillations (HFOs) in human intracereberal EEG

Discrete high-frequency oscillations (HFOs) in the range of 80–500 Hz have previously been recorded from human epileptic brains using intracereberal EEG and seem to be a reliable biomarker of seizure onset zone in patients with intractable epilepsy. Visual marking of HFOs bursts is tedious, highly time-consuming particularly for analyzing long-term multichannel EEG recordings, inevitably subjective and can be error prone. Thus, the development of automatic, fast and robust detectors is necessary and crucial for HFOs investigation and for propelling their eventual clinical applications. This paper presents a proposed algorithm for detection and classification of HFOs, which is a combination of both smoothed Hilbert Huang Transform (HHT) and root mean square (RMS) feature. Performance evaluation in terms of sensitivity and false discovery rate (FDR) were respectively 90.72% and 8.23% related to process validation. Indeed, the proposed method was efficient in terms of high sensitivity in which the majority of HFOs visually identified by experienced reviewers was correctly detected, and had a lower FDR. This would mean that only a low rate of detected events was misclassified as candidate HFOs events. The presented software is extremely fast, suitable and can be considered a valuable clinical tool for HFOs investigation.