High-Frequency Oscillations in the Scalp Electroencephalogram: Mission Impossible without Computational Intelligence

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• High-frequency oscillations (HFOs) in electroencephalogram (EEG) show promise as biomarkers for epileptogenicity.
• Automated detection algorithms exist for HFOs in invasive EEG recordings.
• Invasive EEG recordings carry risks and costs, limiting their widespread applicability.

Purpose of the Study:

• To review the challenges and necessity of developing automated detection algorithms for HFOs in scalp EEG.
• To highlight the limitations of existing algorithms when applied to scalp EEG data.

Main Methods:

• Review of existing literature on HFO detection algorithms.
• Analysis of the characteristics of scalp EEG signals compared to invasive EEG.
• Discussion of the impact of low signal-to-noise ratio, artifacts, and physiological activity on HFO detection in scalp EEG.

Main Results:

• Existing automated HFO detection algorithms, developed for invasive EEG, perform unsatisfactorily on scalp EEG.
• Scalp EEG presents significant challenges for HFO detection due to its lower signal-to-noise ratio and presence of artifacts and physiological noise.
• Visual detection of HFOs in scalp EEG is time-consuming and subjective.

Conclusions:

• Automated detection algorithms for HFOs in scalp EEG are highly warranted.
• Development of robust algorithms tailored for scalp EEG is crucial for advancing epilepsy research and clinical practice.
• Overcoming the challenges of scalp EEG signal quality is essential for reliable HFO analysis.