High frequency oscillations (HFO) in scalp EEG are a new and promising epilepsy biomarker. HFO analysis is typically restricted to random and relatively brief sleep segments. However, considerable fluctuations of HFO rates have been observed over the recording nights, particularly in relation to sleep stages and cycles. Here, we identify the timing within the sleep period and the minimal data interval length that allow for sensitive and reproducible detection of scalp HFO. We selected 16 seizure-free whole-night scalp EEG recordings of children and adolescents with focal lesional epilepsy (median age 7.6 y, range 2.2-17.4 y). We used an automated and clinically validated HFO detector to determine HFO rates (80-250 Hz) in bipolar channels. To identify significant variability over different NREM sleep stages and over time spent in sleep, we modelled HFO rate as a Poisson process. We analysed the test-retest reliability to evaluate the reproducibility of HFO detection across recording intervals. Scalp HFO rates were higher in N3 than in N2 sleep and highest in the first sleep cycle, decreasing with time spent in sleep. In N3 sleep, the median reliability of HFO detection increased from 67% to 79% to 100% for 5-, 10-, and 15-min data intervals, improving significantly (p=0.004) from 5 to 10 min but not from 10 to 15 min. In this analysis of whole-night scalp EEG, we identified the first N3 sleep stage as the most sensitive time window for HFO rate detection. N3 data intervals of 10 min duration are required and sufficient for reliable measurements of HFO rates. Our study provides a robust and reliable framework for implementing scalp HFO as an EEG biomarker in pediatric epilepsy.
Variation of scalp EEG high frequency oscillation rate with sleep stage and time spent in sleep in patients with pediatric epilepsy
