070 Respiratory, cardiac, EEG, BOLD signals and functional connectivity over multiple microsleep episodes

Abstract

Abstract Introduction Brief intrusions of unintended sleep can occur in various contexts, for example during resting-state fMRI scans. In addition to changes in neural activity, such microsleep episodes are also associated with shifts in respiration and heartrate. Here we investigated how these concurrent changes alter the dynamics of the BOLD signal in the brain and estimates of functional connectivity. Methods Ten participants underwent 6 runs of 20 minute resting-state fMRI scans with concurrent respiration, PPG and EEG recording. Realtime eye-closure monitoring combined with post eye-opening self-reports were used to identify microsleep episodes of different durations. Results During microsleep, sustained reductions were observed in arousal as assessed by EEG (ratio of alpha to delta and theta bands), as expected. In comparison, cortical BOLD signal exhibited more complex, temporally multiphasic changes which were consistent across different microsleep durations from 4 to 44s: (i) an initial sleep-onset dip reaching a nadir after ~6s, followed by (ii) an increase above wake baseline that plateaued till awakening. On awakening, (iii) a transient positive bump occurred up to 6s, followed by (iv) an undershoot below baseline lasting ~30s. While seen across the whole brain, these changes showed regional variations, e.g., the signal plateau in the thalamus remained below wake baseline. Sleep onset and awakening were also associated with respective reductions and increases in respiration and heart rate, which affect blood oxygen levels. Brain functional connectivity estimates were altered by the frequency of falling asleep, and this was not resolved by global signal regression. Conclusion Falling asleep and awakening are shown here to be associated with large, widespread BOLD signal changes consistent across varied durations of microsleep. These signal changes are intimately intertwined with shifts in respiration and heart rate, which are influenced by common brainstem nuclei controlling sleep. These autonomic contributions to ‘brain signal’ changes at microsleep onset and awakening are integral to sleep, and urge the integration of autonomic and central nervous system contributions to BOLD signal into frameworks for understanding brain function using fMRI. In addition, the correlation between frequency of microsleep and extent of altered functional connectivity highlight the need to minimize sleep during resting state scans. Support (if any) NMRC/STaR/015/2013