Human intracranial high frequency oscillations (HFOs) detected by automatic time-frequency analysis
Sergey Burnos1, Peter Hilfiker2, Oguzkan Sürücü3
1Neurosurgery Department, University Hospital Zurich, Zurich, Switzerland; Institute of Neuroinformatics, ETH Zurich, Zurich, Switzerland.
Plos One
|April 12, 2014
Summary
A new method accurately detects high frequency oscillations (HFOs), potential biomarkers for epilepsy. This approach analyzes time-frequency data, improving HFO detection and showing promise for clinical diagnosis.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Epileptology
Background:
- High frequency oscillations (HFOs) are emerging biomarkers for identifying epileptogenic tissue.
- Current methods for HFO detection require further refinement for clinical application.
Purpose of the Study:
- To develop and validate a novel method for detecting high frequency oscillations (HFOs) in intracranial EEG (iEEG).
- To assess the performance of the new HFO detection method against the seizure onset zone (SOZ).
Main Methods:
- A two-stage detection process was employed: initial event identification by energy and duration thresholds, followed by time-frequency analysis using Stockwell transformation.
- Parameters for HFO recognition were optimized and validated across multiple patient iEEGs.
- HFO areas were defined by HFO rates and compared with the gold standard SOZ.
Main Results:
- The developed detector successfully differentiated HFOs from artifacts and interictal epileptiform spikes.
- HFOs were detected with significant power outside the conventional 80-500 Hz range.
- The HFO area demonstrated high specificity (>90%) in overlapping with the SOZ for most patients, leading to a re-operation in one case.
Conclusions:
- Time-frequency domain analysis significantly improves HFO detection accuracy by reducing spurious signals compared to traditional filtered methods.
- The proposed HFO detection method offers fast computation and promising diagnostic value for clinical use.
