|
VI Spoormaker, R Wehrle, D Hoehn, H Peter, F Holsboer, M Czisch, PG Sämann
Introduction Model-free analysis of fMRI has revealed the existence of several slowly oscillating (< 0.1 Hz) resting networks with particular neuroanatomical and functional properties. The task-negative default mode network has received most research attention, but additional networks consist of sensory networks (e.g. auditory, lateral and medial visual cortex networks), a sensori-motor network, and networks associated with attention and (working) memory (e.g. temporal / lateralized networks). Aim was to track these networks into light and deep sleep in a combined EEG/fMRI study. Methods We analyzed a total of 92 sleep epochs (26, 24, 24 and 18 epochs of wakefulness, sleep stage 1 , sleep stage 2, and slow wave sleep, of each 5 minutes from 25 young, healthy subjects that underwent fMRI/EEG during falling asleep / sleeping (recordings of 27 minutes per session, EPI sequence, 1.5 Tesla, TR 2 s). Functional data were slice-time corrected and realigned with SPM5 (Wellcome Trust Centre for Neuroimaging, London) and further preprocessed with FSL (Oxford FMRIB Centre) and entered into an independent component analysis (ICA). ICA group output was used to back-reconstruct the individual networks, which were subsequently used in a masked analysis extracting individual time-courses of specific networks. These were fast fourier transformed to focus on low frequency-bands of resting networks (0.01 - 0.1 Hz) and tested for differences among sleep stages in a mixed model analysis of variance in SPSS. Results & Discussion Several (but not all) of the classic resting networks continued to oscillate in light and deep sleep. Differential properties such as anatomic and amplitude changes in fluctuations and their implications will be discussed. |
|