Avalanches of brain co-activation during sleep and wakefulness

BOCACCIO H.; SÁNCHEZ S. M.; DE PINO G.; CASTRO M. N.; LAUFS H.; VILLARREAL M.; TAGLIAZUCCHI E.

Buenos Aires

Congreso; Frontiers in Bioscience 3; 2018

IBioBA

In the study of consciousness, human sleep provides a relatively easy access data to explore the neural changes that underlie the reduction of conscious awareness. For this purpose, we study Resting State fMRI data for different non-rapid eye movement sleep stages (Awake, N1, N2, N3) to find differences in large-scale brain dynamics between this altered states of consciousness. We define clusters of contiguous regions that co-activate by a point-process approach (i.e. binarizing the supra-threshold activity). The distribution of cluster sizes follows a power law, reflecting the scale-free properties associated with the critical regime. These distributions show a non-zero probability of finding very large clusters, defining avalanches of brain co-activation. We estimate the slopes of power laws using Maximum Likelihood Estimators (MLE) and made statistical comparisons between stages. We found that the probability of large clusters is significantly higher in N2 than in wakefulness. The same analysis was performed for down-sampled data and for different thresholds of binarization, showing the consistency of the obtained results. We also apply these methods for phase-shuffled data, as a model without correlations, finding the non-triviality of these results. This suggests that there are differences in the co-activation properties that allow to detect the transition between wakefulness and deep sleep.