Abstracts

Rationale: While the scalp EEG during sleep in healthy individuals has been thoroughly studied, the knowledge on physiological intracranial EEG activity is scarce. This multicenter study aims to provide an atlas of normal intracranial EEG during sleep, complementing published results for wakefulness (Frauscher et al., 2018. Brain, 141:1130-44). Methods: Subjects consisted of epileptic patients who underwent intracranial EEG evaluations for drug-resistant focal seizures. EEGs of presumably normal physiological activity from three tertiary epilepsy centers were analyzed. Selected channels were located in non-lesional tissue, outside the seizure onset zone, had no interictal epileptic discharges, and no obvious slow wave anomaly. All contacts were localized in a common stereotactic space to assess the cerebral structure to which each contact belongs, enabling an analysis of EEG activity across subjects. Sleep was scored with scalp EEG, EOG, and/or EMG. During stages N2 and N3 we automatically detected slow waves and spindles; during REM sleep we computed the proportion of delta power minus proportion of beta and gamma power with respect to total power. Results: We identified 10-minute recordings during stages N2 and N3 from 1481 channels with normal physiological activity from 91 patients  (Figure 1), and 1-minute recordings from 1028 channels (65 patients) during REM sleep.The median (5th and 95th percentiles) rate of slow waves was 5.5/min (1.4–15.0/min) during N2 and 5.5/min (2.0–21.1/min) during N3. Figure 1 shows the rates of sleep slow waves in different regions. A global increase in rate of slow waves during N3 is clearly visible. The mean duration (standard deviation) of the sleep slow waves, was 419 ms (132 ms) during N2 and 440 ms (138 ms) during N3, corresponding to frequencies of 1.30 Hz and 1.24 Hz, respectively. The median spindle rate was 0.80/min (0.1–2.7/min) during N2 and 0.53/min (0–2.4/min) during N3. The mean frequency of the spindles during N2 was 11.9 Hz (1.4 Hz), and during N3 it was 11.6 Hz (1.4 Hz). Figure 2 shows the median spindle rate and mean frequency of spindles for different regions. Higher frequencies were observed in parietal and central regions compared to frontal regions. Spindles in the hippocampus had a relatively high frequency. The rate of spindles detected in the occipital and temporal regions was low, as was the frequency. During REM sleep we observed a clear gradient in slowing: less (<0.2) in the anterior frontal region, increasing in the fronto-basal and central regions (0.25-0.35), and increasing further in the parietal (0.35-0.4), temporal (0.35-0.45) and occipital (0.4-0.5) lobes. The amygdala and hippocampus also exhibited high values (0.5-0.6). Conclusions: This atlas of normal intracranial EEG activity during sleep in a common stereotactic space enables to perform direct analysis of EEG activity across subjects using quantitative analysis. This atlas can serve to enhance the interpretation of the abnormal intracranial EEG on the basis of a solid knowledge of the normal EEG, as has been done for scalp EEG. The data is available online on our institute’s webpage (https://mni-open-ieegatlas.research.mcgill.ca) for use by the community. Funding: This work was supported by the Savoy Epilepsy Foundation (grants to B.F. in 2016 and 2018), the Fonds de recherche Santé Québec (salary support to B.F. 2018-21), and the Canadian Institutes of Health Research (grant FDN-143208 to J.G.).