Data from: Human brain-wide activation of sleep rhythms
Data files
Jul 31, 2026 version files 11.01 MB
-
README.md
3.03 KB
-
SO_main_cmap_tmap.nii.gz
1.84 MB
-
SO_main_cmap_zmap.nii.gz
1.84 MB
-
SO_spindle_cmap_tmap.nii.gz
1.84 MB
-
SO_spindle_cmap_zmap.nii.gz
1.84 MB
-
SP_main_cmap_tmap.nii.gz
1.83 MB
-
SP_main_cmap_zmap.nii.gz
1.83 MB
Abstract
During sleep, our brain undergoes highly synchronized activity, orchestrated by distinct neural rhythms. Little is known about the associated brain activation during these sleep rhythms, and even less about their functional implications. In this study, we investigated the brain-wide activation underlying human sleep rhythms by employing simultaneous electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) in 107 participants during nocturnal nap (first half of the night). We identified robust coupling between slow oscillations (SOs) and fast spindles during deep non-rapid eye movement (NREM) sleep (N2/3 stages), with spindle peaks consistently occurring just before the SO UP-state. This SO-spindle coupling was linked to elevated activation in both the thalamus and hippocampus, alongside increased functional connectivity from the hippocampus to the thalamus and from the thalamus to the medial prefrontal cortex (mPFC). An open-ended cognitive state decoding analysis suggested that these activations may relate to episodic memory processes, yet were distinct from task-related networks. Together, these findings highlight the thalamus as a key coordinator of hippocampal-cortical communication during sleep and provide new insights into the mechanisms by which synchronized sleep rhythms may support memory consolidation.
Dataset DOI: 10.5061/dryad.2fqz612x0
Article DOI: 10.7554/eLife.103956.3
Description of the data and file structure
Simultaneous 64-channel EEG and whole-brain fMRI were recorded during the first half of the night in 138 healthy young adults (107 retained after quality control), with a mean in-scanner sleep duration of ~3 hours. The dataset was collected to examine brain-wide activation and connectivity associated with slow oscillations, sleep spindles, and their coupling during NREM sleep.
Files and variables
File: SO_main_cmap_tmap.nii.gz
Description: Group-level (n = 107) t-map for the slow oscillation (SO) main effect, from a one-sample t-test on participant-level contrast estimates. MNI152NLin2009cAsym space; voxel values are t statistics (df = 106), positive indicating greater BOLD signal associated with SO events.
File: SO_main_cmap_zmap.nii.gz
Description: The same SO main-effect map expressed as z scores rather than t statistics.
File: SO_spindle_cmap_tmap.nii.gz
Description: Group-level (n = 107) t-map for SO–spindle coupling (coupled versus non-coupled events), otherwise as above.
File: SO_spindle_cmap_zmap.nii.gz
Description: The same SO–spindle coupling map expressed as z scores.
File: SP_main_cmap_zmap.nii.gz
Description: Group-level (n = 107) t-map for the spindle main effect, otherwise as above.
File: SP_main_cmap_tmap.nii.gz
Description: The same spindle main-effect map expressed as z scores.
Code/software
The data are standard NIfTI-1 volumes and can be opened with any free neuroimaging software, for example nibabel/nilearn (Python), FSL, AFNI, or MRIcroGL. No custom software is required to view the files, and no code or scripts are included with this submission.
The processing pipeline that generated these maps — including EEG preprocessing, sleep staging, SO and spindle detection, fMRI preprocessing, and the first- and second-level GLM — is described in full, with software names and version numbers, in the Methods section of the associated article.
Human subjects data
All participants provided written informed consent prior to participation, including explicit consent for the deposition and public release of their de-identified EEG and MRI data. The study protocol and consent procedures were approved by the Peking University ethics committee (#2015-26, #2018-11-05).
Data were de-identified before deposition as follows. All direct identifiers were removed and each participant was assigned a randomly generated study code; the linking key was retained only by the study team and is not distributed with the data. EEG recordings were exported with all subject-identifying fields and recording dates cleared from the BrainVision header and marker files. No audio, video, photographic, or free-text clinical records were collected or are included in the shared dataset.
