@article{ref1,
title="RoWDI: rolling window detection of sleep intrusions in the awake brain using fMRI",
journal="Journal of neural engineering",
year="2021",
author="Poudel, Govinda Raj and Hawes, Stephanie and Innes, Carrie R. H. and Parsons, Nicholas and Drummond, Sean P. A. and Caeyensberghs, Karen and Jones, Richard D.",
volume="ePub",
number="ePub",
pages="ePub-ePub",
abstract="OBJECTIVE: Brief episodes of sleep can intrude into the awake human brain due to lack of sleep or fatigue-compromising the safety of critical daily tasks (i.e., driving). These intrusions can also introduce artefactual activity within functional magnetic resonance imaging (fMRI) experiments, prompting the need for an objective and effective method of removing these intrusions. APPROACH: We implemented a method to track sleep-like events in the awake human brain via rolling window detection of intrusions (RoWDI) of fMRI signal. These events can then be used in a voxel-wise, event-related analysis of fMRI data. To test this approach, we generated a template of fMRI activity associated with transition into sleep via simultaneous fMRI and EEG (N=10). RoWDI was then used to identify intrusions of this fMRI activity in 20 individuals performing a cognitive task after a night of partial sleep deprivation. RoWDI was further validated in another independent resting-state fMRI dataset (N=56). MAIN RESULTS: Our method (RoWDI) was able to detect frequent sleep-like events during cognitive task performed after sleep deprivation. These sleep-like events were associated with 20% reduction in pupil size and prolonged response time. The BOLD activity during the sleep-like events covered thalami-cortical regions, which although spatially distinct, co-existed with, task-related activity. These key findings were validated in the independent dataset. SIGNIFICANCE: RoWDI can reliably infer spontaneous sleep-like activity in the human brain. Thus, it may also be used as a tool to delineate and account for neural activity associated with wake-sleep transitions in both resting-state and task-related fMRI studies.<p /> <p>Language: en</p>",
language="en",
issn="1741-2560",
doi="10.1088/1741-2552/ac2bb9",
url="http://dx.doi.org/10.1088/1741-2552/ac2bb9"
}