This study was designed as a cross-sectional observational investigation conducted at the Xijing Hospital affiliated with the Air Force Medical University (Xi’an, Shaanxi, China). The research spanned one year, from January 2022 to December 2022. This study primarily recruited undergraduate and postgraduate student volunteers from various colleges in Xi’an, China. The inclusion criteria were: (1) right-handedness; (2) aged 18–30 years; (3) normal intelligence level; and (4) no contraindications for magnetic resonance imaging (MRI) scanning. Participants were excluded if they had a history of alcohol or drug abuse; any neurological condition, mental illness, or sleep disorder; or if they were an extreme “morning type” or “evening type” according to the Munich Chronotype Scale [22]. All participants were asked to record their sleep and wake times on weekdays and weekends. In total, 54 healthy volunteers, consisting of 29 male and 25 female participants with an average age of 22.46 $\pm$ 1.81 years, were recruited for this study. All participants were briefed about the purpose and significance of the study and signed an informed consent form. They were also instructed to refrain from consuming caffeine or alcohol for 1 week prior to the MRI scan. Additionally, all participants scored 5 points on the Pittsburgh Sleep Quality Index, indicating good sleep quality. This study was conducted according to the Declaration of Helsinki and received approval from the Clinical Trial Ethics Committee of the Xijing Hospital (Approval No.: KY20183005-1).

Participants were required to visit the laboratory twice. The first visit, which took place 1 week before the trial, involved a brief introduction to the experimental procedure and completion of the informed consent process. Participants were also given a wristwatch (Mini Mitter; Philips Respironics, Murrysville, PA, USA) to monitor their sleep patterns over the following week, during which they were asked to maintain their normal sleep habits. During the second visit, volunteers reported to the laboratory no later than 6:00 PM to start the trial. During the trial, participants could read, watch TV, or surf the Internet, but they were not permitted to engage in strenuous activities or leave the laboratory. Two research assistants ensured participants did not fall asleep during the trial. MRI scans were conducted every 2 hours from 10:00 PM to 6:00 AM (T${}_1$, 10:00 PM; T${}_2$, 12:00 PM; T${}_3$, 02:00 AM; T${}_4$, 04:00 AM; T${}_5$, 06:00 AM); resting-state fMRI data were collected during each scan. Each participant underwent five scans. Heart rate and respiratory rate were monitored during the MRI scans to confirm that participants remained awake. After each scan, participants completed the Stanford Sleepiness Scale (SSS), selecting the statement (among seven) that best described their current state of alertness. During the daytime prior to the sleep deprivation, participants were instructed to maintain their usual daily routine, with certain stipulations. They refrained from extreme physical activities, daytime napping, and the consumption of caffeine or alcohol. Additionally, they were advised to avoid heavy meals close to bedtime and limit their screen exposure to reduce potential blue light interference with their circadian rhythms. To ensure adherence and consistency, participants documented their day’s activities in a brief questionnaire, which provided us with a controlled overview of their pre-experimental behaviors.

MRI data were collected using a Discovery MR750 3.0-T superconducting magnetic resonance scanner (GE, Boston, MA, USA) equipped with an 8-channel head coil. Rubber pads and earplugs were provided to mitigate noise during the scan. Participants were instructed to keep their eyes open throughout the scan. Research assistants confirmed that participants remained awake throughout the entire process. For resting-state fMRI, a single-excitation spin plane echo imaging sequence was performed. An axial scan was acquired with the scanning angle parallel to the anterior-posterior commissure line. The scanning parameters were as follows: echo time, 30 ms; repetition time, 2000 ms; slice thickness, 3.0 mm; slices, n = 36; scanning mode, continuous with no interval; field of view, 240 mm $\times$ 240 mm; flip angle, 90${}^{\circ }$; and matrix, 128 $\times$ 128. The total acquisition time was 7 minutes, during which 210 images were collected. To ensure that all participants were free of organic lesions, fluid-attenuated inversion recovery T2-weighted, T2-weighted, and diffusion-weighted MRI images were evaluated by three senior radiologists.

In the preprocessing stage of our imaging data, scans were thoroughly reviewed by experienced imaging specialists. They ensured that there were no evident structural abnormalities or overt issues in any of the brain regions of our subjects. Pre-processing of fMRI data was performed using the Data Processing & Analysis of Brain Imaging (http://rfmri.org/DPABI) and Statistical Parametric Mapping 12 (SPM 12; https://www.fil.ion.ucl.ac.uk/spm/) toolboxes in the MATLAB (ver. 2016b; MathWorks, Natick, MA, USA) environment. The pre-processing steps included: (1) conversion of the original Digital Imaging and Communications in Medicine (DICOM) data into an analyzable Neuroimaging Informatics Technology Initiative (NIFTI) format; (2) removal of fMRI data acquired at the first 10 time points for each participant; (3) time slice correction to eliminate interslice signal differences caused by varying acquisition times; (4) motion correction to exclude participants with translational movement $>$3 mm and/or rotational movement $>$3${}^{\circ }$ during data acquisition, and calculation of the average framewise displacement for each participant for use as a covariate in the subsequent statistical analysis; (5) spatial normalization to align individual fMRI images to Montreal Neurological Institute standard space and resampling of the aligned images to 3 mm $\times$ 3 mm $\times$ 3 mm; (6) regression analysis using a general linear model with white matter, cerebrospinal fluid, and Friston’s 24 head motion parameters as covariates; (7) bandpass filtering (0.01–0.10 Hz) and detrending to reduce interference by irrelevant noise; and (8) spatial smoothing using a 6 mm $\times$ 6 mm $\times$ 6 mm full-width at half-maximum Gaussian kernel filter.

FCD is computed through a series of steps. First, whole-brain voxel-wise correlation is performed to identify all voxels that show a significant temporal correlation with the seed voxel. The threshold for significance is set at a correlation coefficient of 0.6. The number of significantly correlated voxels is then determined to generate the global FCD (gFCD) value for the seed voxel. This process is repeated for every voxel in the brain, producing a 3D map of gFCD. The short-range functional connectivity density index (iFCD) is derived in a similar way to the gFCD value, but connections are counted only if significantly correlated voxels are located within a radius of 12 mm. The gFCD reflects the strength and extent of functional connections between spatially distant or interregional brain regions; in contrast, the iFCD provides an index of the degree of local integration within the brain network, reflecting the strength and extent of functional connections between neighboring regions. Finally, normalized FCD values are computed for subsequent statistical analysis. For our study, we adopted a whole-brain approach to ensure comprehensive coverage and understanding of potential neural changes. This meant that we did not limit our analysis to predefined regions but investigated the entirety of the brain to detect any alterations in functional connectivity.

Demographic data were statistically analyzed using SPSS software (version 24.0; IBM Corp., Armonk, NY, USA). Given the longitudinal nature of the experiment, a one-way repeated-measures analysis of variance (ANOVA) was used to compare differences in SSS values at multiple time points. Post hoc analysis further explored differences in SSS values between pairs of time points. False discovery rate (FDR)-corrected p-values 0.05 were considered indicative of significant differences. Measurement data conforming to a normal distribution are presented as mean $\pm$ standard deviation. Statistical parametric mapping software (SPM 12, London, UK, https://www.fil.ion.ucl.ac.uk/spm/) was employed to analyze the gFCD and iFCD values of the entire brain at multiple time points. Initially, using the factor design module of SPM 12 software, a one-way-repeated measures ANOVA was used to compare brain regions with differing gFCD and iFCD values at multiple time points based on the default grey matter template. The FDR method was applied to correct for multiple comparisons. Subsequently, the gFCD and iFCD values of various brain regions were extracted at each time point for post hoc analysis. A Pearson correlation analysis was performed to determine the correlations between gFCD/iFCD and SSS values within various brain regions. Bonferroni-corrected p-values 0.05 were considered indicative of a significant difference.

A one-way repeated-measures ANOVA indicated significant changes in gFCD values in the frontal gyrus (right middle, right superior, and left medial), parietal gyrus (bilateral superior), temporal gyrus (left superior and left middle), occipital gyrus (right superior), left precentral gyrus, bilateral thalamus, right middle cingulate gyrus, and right angular and left supramarginal gyrus (Table 1). Significant changes in iFCD values were mainly found in the frontal gyrus (right middle, right superior and left medial), parietal gyrus (bilateral superior), temporal gyrus (left superior), bilateral thalamus, bilateral precentral gyrus, and middle temporal gyrus (Table 2). The overall gFCD and iFCD findings are shown in Fig. 2.

Fig. 2.

This figure summarizes the significant changes in global and local functional connectivity density (gFCD and iFCD) in various brain regions during sleep deprivation. Results from a one-way repeated-measures ANOVA are highlighted, showing alterations in frontal, parietal, temporal, and other key brain regions. Specific FCD changes are listed in Table 1 and Table 2. FCD, functional connectivity density.

To investigate the influence of sleepiness on functional connectivity during SD, Pearson correlation analysis was used to evaluate the relationship between sleepiness and gFCD/iFCD values in the brain areas mentioned above. The gFCD values of the left superior temporal gyrus and left thalamus were positively correlated with the SSS scores (r = 0.423, p 0.001, and r = 0.407, p 0.001, respectively). Sleepiness was positively correlated with iFCD values in the left superior temporal gyrus (r = 0.319, p 0.001) and negatively correlated with iFCD values in the right superior frontal gyrus (r = –0.408, p 0.001). Detailed results are visualized in Fig. 3.

Fig. 3.

This figure displays the Pearson correlation analysis between Stanford Sleepiness Scale (SSS) scores and functional connectivity density (FCD) values in selected brain regions. Significant positive correlations were found in the left superior temporal gyrus (a,b) and left thalamus (c), while a negative correlation was found in the right superior frontal gyrus (d). The strength and significance of each correlation are noted.

To reveal detailed dynamic changes in these regions, the gFCD/iFCD values for each region and session were extracted and are displayed in Fig. 4. The gFCD values in the left superior temporal gyrus and left thalamus, as well as the iFCD values in the left superior temporal gyrus, gradually increased across the five sessions, while the iFCD values in the right superior frontal gyrus gradually decreased.

Fig. 4.

This figure shows the dynamic pattern of FCD alterations during sleep deprivation. The gFCD and iFCD in the left superior temporal gyrus and the gFCD in the left thalamus gradually increased, while the gFCD in the right superior frontal gyrus gradually decreased.

In our investigation, we identified significant alterations in global and local functional connectivity, predominantly within the middle/superior temporal gyrus and the medial prefrontal cortex. These core regions of the DMN play pivotal roles in an array of behavioral, cognitive, and physiological activities. We noted a progressive increase in gFCD/iFCD values in the left middle temporal gyrus as sleep deprivation advanced. This uptick in activity within the DMN suggests it could be instrumental in the initiation of drowsiness. Past research, utilizing methodologies like independent component analysis and seed-based correlation analysis, has revealed irregular functional connections within the DMN and between the DMN and other cerebral networks [23]. Building on our prior findings, we observed that during the slowest 20% of reactions in psychomotor vigilance tasks, there was a linear escalation in task-related activation within the DMN [24]. The confluence of these observations underscores the hypothesis that heightened activity in the DMN might be the neural underpinning for the inattention and propensity for mind-wandering exhibited during prolonged wakefulness. Further, these discoveries, set within the broader landscape of DMN research, emphasize the essential role of sleep in modulating this pivotal network and its broader implications for cognitive health and attentional capacities.

In this study, low gFCD/iFCD values were also found in the superior frontal gyrus and superior parietal gyrus, which together comprise the FPN. Functional connectivity in the right superior parietal gyrus decreased linearly during SD, indicating that high-order FPN activities were heavily influenced by circadian rhythms and sleep pressure. Numerous neuroimaging studies have detected abnormal functional activity in the FPN after SD compared with that seen after normal sleep. Drummond et al. [25] observed that the degree of activation within the prefrontal cortex, parietal lobe, and premotor cortex decreased significantly after SD. Similarly, our previous positron emission tomography study revealed that glucose metabolism within the prefrontal cortex, parietal lobe, and thalamus dropped significantly after SD [26]. Moreover, our recent studies indicated that the FPN plays an essential role in maintaining cognitive stability; SD-resistant individuals showed more efficient information transmission than SD-vulnerable individuals [27].

The thalamus, strategically positioned within the central nervous system, is pivotal in governing the transmission of awakening signals from the brain stem and acts as a linchpin in maintaining the arousal state of an organism [28]. It’s no surprise that a multitude of studies spanning disciplines, from physiology to genetics, have highlighted the thalamus’s multifaceted roles in regulating consciousness, orchestrating both sleep and wake states [29]. Our central observation during sleep deprivation concerns the left Mediodorsal Thalamus (MD thalamus). The intricate connectivity of the MD thalamus with the prefrontal cortex is key to various cognitive functions, ranging from attention to executive decision-making. Under sleep deprivation, these connections could be disrupted, potentially leading to observed cognitive deficits, such as reduced attention and executive function capabilities.

The implication of sleep deprivation on the thalamus is profound. An unstable state of thalamic activity might be incited by sleep deprivation. Enhanced thalamic activity ensures effective transmission of ascending arousal signals from the brain stem to the DMN and FPN networks, preserving their intrinsic antagonistic dynamics [30]. Conversely, a decline in this activity could intercept these arousal signals within the thalamus itself, thereby upsetting the equilibrium between the default and attention networks. However, our study ascertained that thalamic activity consistently augmented during SD. Such a surge in activity may symbolize a compensatory mechanism, striving to counteract sleep deprivation’s adverse impacts, and thus aiding individuals in preserving normal behavior and performance. In summary, our results accentuate the quintessential role of the thalamus, especially the MD thalamus, in upholding wakefulness during episodes of sleep deprivation.