Investigating sleep in the aging human brain using fast multimodal neuroimaging
OA Version
Citation
Abstract
Sleep is essential to life across scales and plays a fundamental role in maintaining brain health and cognitive function. As humans we spend approximately 1/3 of our lives sleeping, but as we age there are significant changes to sleep quality and quantity. These changes occur as early as middle age, preceding any neurological or neurodegenerative conditions, and are hypothesized to contribute to the neurodegenerative process. Sleep serves many critical functions including memory consolidation and clearance of metabolic waste products from brain tissue via interactions with cerebrospinal fluid, or CSF, the primary waste transport system of the brain. CSF circulation is particularly heightened during sleep, and a breakdown in this circulation has been proposed as a key mechanism contributing to the robust relationship between sleep and long-term brain health. Despite these links, it remains unknown whether CSF circulation during sleep decreases in the aging brain prior to disease onset, and, if so, by what mechanisms. Additionally, age-related declines in sleep quality and quantity are hypothesized to causally contribute to the increased incidence and accelerated course of neurodegeneration in the older adult brain. Although altered sleep behavior is common even in normative aging, our understanding of why older adults experience more fragmented and fragile sleep is far from complete. This dissertation addresses these knowledge gaps by first examining whether and how CSF flow during sleep changes in the aging brain, and then investigating the underlying mechanisms behind sleep impairments in older adults.To address these gaps in knowledge we collected a simultaneous electroencephalography (EEG) and fast functional magnetic resonance imaging (fMRI) dataset from young and older adults during nighttime rest. Importantly, we also measured CSF flow into the fourth ventricle to obtain an integrated measure of neural and physiological dynamics in the aging brain during sleep. We first tested whether CSF flow during sleep is altered in the aging brain and found a significant reduction in low-frequency, pulsatile CSF flow in older adults, driven by age-related changes in the neural and vascular drivers of CSF flow. We next investigated age-related changes in whole brain dynamics during sleep. Using an event locked analysis centered around behavioral awakenings, we identified a subset of cortical and subcortical regions with significantly altered activity around arousal, including arousal-promoting regions that display hyperactivity following transitions in behavioral arousal state. We also employed a Gaussian Linear Hidden Markov Model (GLHMM) to characterize age-related changes in spontaneous fluctuations in dynamic functional connectivity (dFC) revealing a sequence of brain states underlying behavioral arousal state transitions – one of which mimics the spatiotemporal pattern of the arousal-locked response. Older adults spend significantly more time in this transition-related state, as well as one characterized by reduced global cortical connectivity. Together these results provide novel insights into sleep in the aging human brain finding altered CSF flow – potentially compromising waste clearance and promoting harmful metabolite buildup – along with alterations in arousal-regulating regions and networks.
This dissertation will also include an exploration of the spectral properties of fast fMRI signals and whether resting-state signals can provide a task-free way of characterizing voxelwise differences in the timing of the hemodynamic response. We show that not only is there rich information about hemodynamic response timing in the resting-state frequency spectrum, but also these features better distinguish the temporal properties of individual voxels compared to the current gold standard breath hold method for characterizing variability in the temporal properties of the hemodynamic response. Ultimately, our findings offer new insights into the temporal properties of fMRI signals across voxels, which is vital for accurate fMRI analyses and enhances fast fMRI’s ability to identify and monitor rapid neural dynamics.
Description
2026
License
Attribution-NonCommercial 4.0 International