Exploration of functional brain network during sleep and wakefulness using simultaneous EEG and fMRI

Abstract

Sleep is regulated by modulation of neural interaction between particular brain regions. Brain neural activity can be estimated by various neuroimaging methods such as electroencephalography (EEG), functional magnetic resonance imaging (fMRI) and so on. Further analysis of the acquired data based on graph theoretical approaches aims to identify the functional brain networks. Moreover, in the recent neuroscience field, many research studies have defined the brain as a large complex system that is organized by the reciprocal interactions between brain regions.To investigate the changes of functional brain network during sleep and wakefulness (awake), we acquired the data from 11 healthy participants using simultaneous EEG and fMRI. The sleep data were obtained approximately 7 hours while participants are sleeping. Unfortunately, the artifacts caused by movements of participants as well as characteristics of instruments had occurred inevitably during the process of data acquisition. Thus we removed these artifacts using the average artifact subtraction for EEG data and the sub-volume utilization for fMRI data. Then, the preprocessed sleep data were visually classified and scored for every 30 sec epochs by sleep S1, S2, S3 and REM according to the international standard of AASM (American Academy of Sleep Medicine) guideline. In network analysis, to construct the functional brain network, we defined nodes depicting 95 regions of interest using brain anatomical atlas from the Automated Anatomical Labeling (AAL) and Freesurfer. For the construction of functional connectivity we applied the Pearson’s correlation coefficient between pairs of node time series.The ventral diencephalon (vDC) seeded functional connectivity has dense connections in the sleep associated regions with subcortex during sleep stages rather than wakefulness. Comparing the network topology of whole

brain, we observed the main effect of sleep in the mean clustering coefficient that was significantly increased during sleep S2, S3 and REM compared with wakefulness. The global / local efficiency showed significant differences between the pairs of only sleep S1, the transitional state from wake to sleep. This functional brain network could be decomposed with independent functional subnetworks revealed by graph ICA (independent component analysis). Our results support the assumption that the functional brain network is organized by locally distributed unit functions and it is functionally segregated during sleep.In this study, we observed that the sleep data could be acquired by concordant EEG and fMRI during very long time. Our result demonstrated that the functional brain network could be reorganized by the dynamic changes of spontaneous brain activity depending on sleep stages.