Effect of sampling rate and filter settings on High Frequency Oscillation detections

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• High Frequency Oscillations (HFOs) are investigated as potential biomarkers for epilepsy.
• Variability in data acquisition parameters across centers poses challenges for consistent HFO detection and analysis.

Purpose of the Study:

• To quantify the effects of sampling rate (FS) and anti-aliasing filter (AAF) settings on automated HFO detection.
• To evaluate the impact of these parameters on HFO sensitivity, features, and the identification of the seizure onset zone.

Main Methods:

• Intracranial EEG data from 17 epilepsy patients recorded at 5kHz FS were analyzed.
• Data were downsampled and filtered to simulate various acquisition settings.
• HFO detection sensitivity, feature distributions, and seizure onset zone identification were compared across simulated parameters.

Main Results:

• Relative HFO detection sensitivity remained above 80% with FS ≥2kHz or AAF ≥500Hz.
• HFO feature distributions were consistent down to 1kHz FS or 200Hz AAF (AUROC<0.7).
• HFO rate effectively identified the seizure onset zone, but peak frequency was highly variable across settings.

Conclusions:

• A minimum sampling rate of 2kHz and AAF of 500Hz are recommended for robust HFO detection.
• HFO peak frequency is not a reliable feature due to variability under different acquisition settings.
• These findings provide essential guidance for clinical centers standardizing HFO analysis.