Laboratorio de Sueño y Cronobiología, Programa de Fisiología y Biofísica, Instituto de Ciencias Biomédicas (ICBM), Facultad de Medicina, Universidad de Chile, Santiago, Chile.
Laboratorio de Sueño y Cronobiología, Programa de Fisiología y Biofísica, Instituto de Ciencias Biomédicas (ICBM), Facultad de Medicina, Universidad de Chile, Santiago, Chile.
Neuroimage. 2018 May 15;172:575-585. doi: 10.1016/j.neuroimage.2018.01.063. Epub 2018 Feb 2.
Traditionally, EEG is understood as originating from the synchronous activation of neuronal populations that generate rhythmic oscillations in specific frequency bands. Recently, new neuronal dynamics regimes have been identified (e.g. neuronal avalanches) characterized by irregular or arrhythmic activity. In addition, it is starting to be acknowledged that broadband properties of EEG spectrum (following a 1/f law) are tightly linked to brain function. Nevertheless, there is still no theoretical framework accommodating the coexistence of these two EEG phenomenologies: rhythmic/narrowband and arrhythmic/broadband. To address this problem, we present a new framework for EEG analysis based on the relation between the Gaussianity and the envelope of a given signal. EEG Gaussianity is a relevant assessment because if EEG emerges from the superposition of uncorrelated sources, it should exhibit properties of a Gaussian process, otherwise, as in the case of neural synchronization, deviations from Gaussianity should be observed. We use analytical results demonstrating that the coefficient of variation of the envelope (CVE) of Gaussian noise (or any of its filtered sub-bands) is the constant 4π-1≈0.523, thus enabling CVE to be a useful metric to assess EEG Gaussianity. Furthermore, a new and highly informative analysis space (envelope characterization space) is generated by combining the CVE and the envelope average amplitude. We use this space to analyze rat EEG recordings during sleep-wake cycles. Our results show that delta, theta and sigma bands approach Gaussianity at the lowest EEG amplitudes while exhibiting significant deviations at high EEG amplitudes. Deviations to low-CVE appeared prominently during REM sleep, associated with theta rhythm, a regime consistent with the dynamics shown by the synchronization of weakly coupled oscillators. On the other hand, deviations to high-CVE, appearing mostly during NREM sleep associated with EEG phasic activity and high-amplitude Gaussian waves, can be interpreted as the arrhythmic superposition of transient neural synchronization events. These two different manifestations of neural synchrony (low-CVE/high-CVE) explain the well-known spectral differences between REM and NREM sleep, while also illuminating the origin of the EEG 1/f spectrum.
传统上，脑电图被理解为源自同步激活神经元群体，这些神经元群体在特定频率带中产生节律性振荡。最近，已经确定了新的神经元动力学状态（例如神经元瀑流），其特征是不规则或无节律性活动。此外，人们开始认识到脑电图频谱的宽带特性（遵循 1/f 定律）与大脑功能紧密相关。然而，目前仍然没有理论框架来容纳这两种脑电图现象学的共存：节律性/窄带和无节律性/宽带。为了解决这个问题，我们提出了一种基于给定信号的高斯性和包络之间关系的新脑电图分析框架。脑电图的高斯性是一个相关的评估，因为如果脑电图是从不相关源的叠加中产生的，它应该表现出高斯过程的性质，否则，就像在神经同步的情况下，应该观察到偏离高斯性。我们使用分析结果证明，高斯噪声（或其任何滤波子带）的包络的变异系数（CVE）是常数 4π-1≈0.523，因此 CVE 可以作为评估脑电图高斯性的有用指标。此外，通过组合 CVE 和包络平均幅度，生成了一个新的、信息量非常丰富的分析空间（包络特征空间）。我们使用这个空间来分析大鼠在睡眠-觉醒周期中的脑电图记录。我们的结果表明，在最低的脑电图幅度下，delta、theta 和 sigma 波段接近高斯性，而在高脑电图幅度下则表现出显著的偏差。在 REM 睡眠期间，出现了低 CVE 的偏差，与 theta 节律相关，这一状态与弱耦合振荡器同步的动力学一致。另一方面，在 NREM 睡眠期间，主要与 EEG 相位活动和高幅度高斯波相关的高 CVE 偏差可以解释为瞬态神经同步事件的无节律叠加。这两种不同的神经同步表现（低 CVE/高 CVE）解释了 REM 和 NREM 睡眠之间众所周知的光谱差异，同时也阐明了 EEG 1/f 光谱的起源。
