Rationale: We recently proposed a novel brain ‘map’ (Kalamangalam et al., Clin Neurophysiol. 2020; 131(3), 665-675) of normal human intracranial EEG (iEEG) in the wake state, using a power spectrum-based modeling approach and normative iEEG data from a publicly available database that we call the wake Montreal Neurological Institute Atlas (wMNIA) (Frauscher et al., Brain 2018; 141, 1130-44). In this work we extend our previous methods to the expanded MNIA, which now includes sleep data (sMNIA, representing stages N2, N3 and REM) (von Ellenrieder et al., Ann. Neurol. 2020; 87(2), 289-301). Methods: The data were downloaded (https://mni-open-ieegatlas.research.mcgill.ca), high-pass filtered above 1 Hz, notch-filtered at 60 Hz, demeaned, normalized to unit variance, and divided into non-overlapping 10s segments to compute the Welch power spectral density in the 1-100 Hz frequency range. The region-specific average or smoothed amplitude spectral densities were resampled on a logarithmic (base 2) frequency scale, and fit to a five-component Gaussian mixture (GM) model of the Berger bands δ (1-4 Hz), θ (4 – 8 Hz), α (8- 16 Hz), β (16 – 32), and γ (32-100). Each Gaussian was defined by its amplitude (A), mean (μ) and standard deviation (σ), yielding a multicomponent model was simplified to five dimensions. Each 5-dimensional vector model was converted into a red-green-blue composite color with intensity and hue variations. The colors were used to create region-specific mosaics on to the ICBM152 cortical surface model across the states of wakefulness, N2 sleep, N2 sleep, and REM sleep. Results: The GM model fits were uniformly excellent (R2 > 0.99) for all w- and sMNIA brain areas. Region-specific trends and commonalities in the spectral model between brain regions and across sleep-wake states were identified. Observations unrelated to the model were that the primary cortical areas of vision, motor function and audition, in addition to the hippocampus, did not appear to participate in the ‘awakening’ of the cortex during REM sleep. Conclusions: Spectral variation in the intracranial EEG is continuous in space (across different cortical regions) and time (stage of circadian cycle), arguing for a ‘continuum’ hypothesis in the generative processes of the iEEG during sleep and wakefulness in human brain. Funding: Please list any funding that was received in support of this abstract.: This work was partially supported by the Wilder family endowments to the University of Florida.
.jpg "Sarah E. Long photo")