Whole-Brain Static Functional Connectivity Disruptions Based on the Default Mode Network in Patients with Mild Cognitive Impairment

Abstract

Mild cognitive impairment (MCI) is regarded a potential early stage of Alzheimer’s disease (AD) and associated with a significantly increased risk of progression to AD. This study aims to evaluate whole-brain static functional connectivity (SFC) disruptions with the default mode network (DMN) seed points in patients with MCI by resting-state functional magnetic resonance imaging (rs-fMRI), and to explore whether these disruptions could serve as potential markers for MCI progression to AD. Retrospective rs-fMRI data with MCI (n = 36) and corresponding matched healthy controls (HCs) (n = 26) were collected for comparison. Independent component analysis (ICA) was used to extract DMN regions, and SFC was calculated for four seed points within the DMN. Two-sample t-tests were performed to compare group differences in SFC strength between the MCI and HC groups, and Pearson correlation analyses were conducted. Compared to HCs, the MCI group showed both increased and decreased SFC between four subregions and multiple brain regions, decreased SFC was more widely distributed than increased SFC. Abnormal connectivity was more prominent in the first two key nodes compared to the latter two. Affected regions primarily located in the precuneus, frontal gyri, temporal gyri, postcentral gyrus, caudate nucleus, lingual gyrus, and fusiform gyrus. The SFC value between the right angular gyrus and the right insula was significantly negatively correlated with MoCA scores (r = − 0.385, p < 0.05, FDR-corrected). It reveals a decline in the functional integration capacity within the DMN, as well as complex reorganization and abnormal connectivity patterns between the DMN and other brain networks. The altered interactions between DMN subregions and abnormal brain areas are significantly associated with episodic memory disturbance in MCI.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10548-026-01187-6.

Introduction

Alzheimer’s disease (AD), an irreversible neurodegenerative disease that often remains undetected, occurs in the elderly. Clinically, AD begins with mild cognitive impairment (MCI), manifesting as an objective decline in cognitive function whereas normal daily life remains relatively normal (Chen et al. 2024). According to epidemiological studies, approximately 10%-15% of MCI patients develop dementia per year (Wang et al. 2021), with over 50% of cases developing AD longitudinally (Gauthier et al. 2006). The etiology of AD is intricate and diverse (Zhang et al. 2024), early interventions may delay or prevent progression of MCI (Wang et al. 2024).

Normal cognitive functions rely on the integrity of large-scale brain networks (Luo et al. 2023, Valera-Bermejo et al. 2021), with the default mode network (DMN) receiving particular attention due to its critical involvement in social cognition and episodic memory—functions that are affected early in the AD continuum. Key anatomical regions of the DMN include the posterior cingulate cortex/precuneus (PCC), medial prefrontal cortex (MPFC), bilateral angular gyri (AG_R/L), bilateral temporal cortices, hippocampal regions, and inferior parietal lobules (Menon 2023).

Due to its convenience, non-invasiveness, non-ionizing radiation, and high spatial resolution, resting-state functional magnetic resonance imaging (rs-fMRI) has emerged as a widely used neuroimaging tool for studying AD and MCI. It is primarily used to explore the functional architecture and spontaneous brain activity, which can serve as a potential approach to reveal abnormalities in functional networks associated with cognitive impairment (Wang et al. 2023). Static functional connectivity (SFC) of the brain network refers to the inter-regional synchrony, reflected by synchronized fluctuations of brain blood oxygen level-dependent (BOLD) signals (Ibrahim et al. 2021). Rs-fMRI has been widely used to demonstrate SFC abnormalities across brain networks in MCI, providing a reliable measure of connectivity during the brain’s resting state (Dang et al. 2023).

Previous studies have shown that pathological amyloid-β deposition affects core regions of the DMN before the onset of clinical symptoms in AD (Koch et al. 2015, Corriveau-Lecavalier et al. 2023). Consistently, a large body of rs-fMRI studies have demonstrated that MCI exhibit distinct SFC abnormalities within the DMN—characterized by reduced functional connectivity between left PCC and left parahippocampal gyrus (Liu et al. 2016), a significant negative correlation between MPFC connectivity with parahippocampus and posterior hippocampus and visual naming performance in MCI (Warren and Moustafa 2023), and a positive association between greater severity of global cognitive impairment in MCI and stronger SFC in the bilateral MPFC (including the anterior cingulate cortices), left angular gyrus (AG), and right temporal pole (Wang et al. 2018). However, previous studies have mainly focused on intra-DMN functional alterations using pre-defined ROIs, which are limited by subjective prior knowledge and may hinder comprehensive capture of global/local functional changes, and lack systematic analysis of SFC patterns between DMN subregions and the whole brain. Therefore, functional alterations in the DMN may serve as potential targets for the assessment and early diagnosis of MCI.

In our team previous findings, lateralization in the distribution of the amplitude of low-frequency fluctuation (ALFF) values were observed between the MCI and HC groups, with the most extensive alterations in local brain activity observed in the DMN (Shao et al. 2025). Therefore, we further hypothesize that DMN SFC abnormalities may represent a potential associative factor underlying the progression from MCI to AD. This study used independent component analysis (ICA) to extract DMN subregions, and rs-fMRI to investigate seed-based whole-brain SFC patterns of the DMN subregions in patients with MCI, aiming to explore whether these abnormalities could serve as potential biomarkers for the conversion from MCI to AD.

Materials and Methods

Subjects

From October 2020 to February 2024, the study consecutively recruited patients with MCI (n = 41) aged between 50 and 85 years from the Department of Neurology or Gerontology of East Hospital, Tongji University. The healthy controls (HCs) group (n = 33), consisting of age-matched, healthy, elderly volunteers, were recruited from the community advertisements. Among them, 5 MCI and 7 HCs were excluded due to excessive head motion (cumulative translation or rotation > 3.0 mm or 3.0°), a total of 36 patients with MCI and 26 healthy controls were included in the final analysis. After registration, the study protocol was reviewed and approved by the Ethics Review Committee of Dongfang Hospital, Tongji University. Informed consent, either written or verbal, was obtained from all participants before their voluntary participation, ensuring they fully understood the study’s objectives, procedures, and safety precautions. Verbal consent was formally documented via a standardized, dual-signed consent form and additionally audio-recorded for verification.

Neuropsychological Assessment

Before the imaging scanning, general details such as name, age, gender, educational level (years), handedness, and present and past medical history were collected. Meanwhile, two senior specialists assessed the cognitive function scale of all subjects independently in a quiet environment. The mini-mental state examination (MMSE) is commonly used in clinical and research settings to assess the severity of dementia (Folstein et al. 1975), with the following cutoff points: MMSE ≤ 17 for illiterate individuals, ≤ 20 for those with primary school education, and ≤ 24 for those with junior high school education or higher (Judge et al. 2019). Montreal cognitive assessment (MoCA), a recommended tool for MCI screening in China (Nasreddine et al. 2005), employs specific cutoff points based on Chinese norms: (Lu et al. 2011)≤ 13 for illiterate individuals, ≤ 19 for individuals with primary school education, and ≤ 24 for those with junior high school education or higher. Due to the low accuracy and sensitivity of MMSE in identifying HCs from patients with MCI, it was used only as a supplementary tool in this study (Jia et al. 2021). Specifically, MMSE scores served auxiliary grouping and descriptive analysis purposes, and were excluded from all inferential statistical analyses.

Inclusion and Exclusion Criteria

MCI group inclusion criteria were as follows: (1) self-reported MCI, with cognitive decline validated by an informant or a clinician and supported by objective evidence; (2) unaffected ability to work and live independently; (3) MMSE and MoCA scores meeting the diagnostic criteria for MCI, as specified in the above Neuropsychological assessment section; (4) right-handedness (to reduce FC hemispheric variability and ensure consistent inter-subject brain network comparisons); (5) cognitive abnormalities not solely attributed to normal aging; and (6) ≥ 6 years of education; (7) included well-controlled chronic diseases such as hypertension, diabetes mellitus without impacts on cognitive function or cerebral hemodynamics.

Inclusion criteria for the HC group were as follows: (1) no self-reported cognitive impairment (including memory decline) and no history of structural neurological or psychiatric disorders; (2) MMSE score ≥ 27 and MoCA score ≥ 26; (3) gender, age, and educational level matched to those of the MCI group; and (4) right-handedness.

Exclusion criteria were as follows: (1) MRI scanning contraindications, such as claustrophobia and metal implants; (2) inability to cooperate with long-term resting state examination, such as critical illness, slurred speech, mental disorders; (3) neurological diseases, such as brain tumors, stroke, cerebral infarction, and traumatic brain injury; (4) use of drugs affecting the nervous system or history of severe mental disorders, such as major depression and generalized anxiety disorder; (5) presence of severe uncontrolled endocrine diseases, such as hyperthyroidism, hypothyroidism, and diabetes mellitus; (6) other neurodegenerative diseases that can cause cognitive dysfunction, such as Parkinson’s disease, meningitis, multiple sclerosis, cerebrovascular seizures, Huntington’s disease, and corticobasal ganglia degeneration.

MRI Acquisition

All MRI images were obtained using an M750w 3.0T GE MRI (GE Healthcare, American) scanner. During scanning, the technician placed the subjects in a supine position, with the head positioned in the center of the 32-channel phased array head coil; the head was fixed with a sponge cushion. Continuous motion monitoring was conducted via visual observation by the technician through the scanner’s monitoring window, with verbal reminders provided promptly if slight head movement was detected. Participants were asked to lie still to minimize involuntary head movements, and cotton balls were placed in their ears to reduce the noise from the MRI machine. All subjects were instructed to keep their eyes closed and remain relaxed and awake.

T2-weighted imaging (T2WI) and fluid-attenuated inversion recovery imaging (FLAIR) were employed to identify brain lesions without atrophy, such as small vascular infarcts, demyelinating lesions, or focal edema. Diffusion-weighted imaging (DWI) was used to rule out acute ischemic lesions or other structural abnormalities, and magnetic resonance angiography (MRA) was used to exclude cerebrovascular lesions. Detailed scanning parameters are shown in Table 1.

Table 1.

Conventional MRI sequence parameters

Parameters	T2WI	FLAIR	DWI	MRA
TR (ms)	4000	8500	13,700	17.5
TE (ms)	112	92.5	85	4
FOV (mm×mm)	230 × 230	230 × 230	224 × 224	220 × 220
Slice thickness/ gap (mm/mm)	5/0.6	5/0.6	2/0.6	1/0

TR: repetition time; TE: echo time; FOV: field of vision

3D high-resolution T1-weighted (3D-T1WI) images were acquired using the 3D-BRAVO sequence to obtain whole-brain sagittal images to provide a template for segmenting ROIs, such as gray matter, white matter, and cerebrospinal fluid. These images were reviewed by two experienced radiologists to rule out any structural abnormalities. BOLD sequences were acquired using echo-planar imaging allowing for rapid imaging with minimal impact on image quality. This method provided whole brain rs-fMRI data for subsequent processing and analysis. The specific imaging parameters are shown in Table 2.

Table 2.

Structural 3D and functional imaging sequence parameters

Parameters	3D T1WI	BOLD-fMRI
TR (ms)	8.5	2000
TE (ms)	3.2	30
Flip angle (°)	7	90
FOV (mm×mm)	256 × 256	192 × 192
Matrix	256 × 256	64 × 64
Slice thickness (mm)	1	3
Slice gap (mm)	0	1
Slices	188	36
Slice time (s)	339	360

Data Processing

To assess image quality and exclude morphological abnormalities, conventional and functional images were first reviewed by two experienced radiologists. Rs-fMRI data were preprocessed and subjected to seed-based SFC analysis using REST-plus (http://www.restfmri.net/) implemented on the MATLAB R2022b platform (https://www.mathworks.cn/). The specific steps were as follows: (1) conversion of all initial DICOM data to NIfTI format; (2) removal of the first five volumes of rs-fMRI data; (3) slice-timing correction; (4) realignment, with participants exhibiting cumulative head motion exceeding 3.0 mm in translation or 3.0° in rotation excluded (5 MCI and 7 HCs were excluded), and framewise displacement (FD) was calculated per timepoint (scans with FD > 0.9 mm were flagged as outliers); (5) spatial normalization to the Montreal Neurological Institute (MNI) standard template; (6) smoothing with an 8 mm full-width at half-maximum Gaussian kernel (preprocessing terminated here for independent component (IC) extraction); (7) detrending and nuisance covariates regression (including Friston-24 head motion parameters, white matter, and cerebrospinal fluid signals; global signal not included to prevent over-elimination of physiological connectivity between brain regions); (8) bandpass filtering (0.01–0.1 Hz).

In this study, IC extraction was performed using the GIFT toolbox: (1) The optimal number of ICs for subject decomposition was estimated using the minimum description length criterion, which was applied at the individual subject level prior to group-level ICA; (2) principal component analysis reduced data dimensions, yielding 165 ICs separated by the post-nonlinear ICA FastICA algorithm; (Bi et al. 2018) this number was consistent across all subjects, and 165 ICs sufficiently cover core functional networks and potential sub-network components while avoiding over-decomposition of noise; and (3) spatial correlation coefficients of each IC with the network template were calculated based on the prior template of resting-state networks (RSNs) provided by Mantini (Mantini et al. 2007, Mantini et al. 2009). ICs with a spatial correlation coefficient threshold of r > 0.3 were further screened, and the ICs with the highest correlation coefficients were identified as the RSNs corresponding to the prior template. Eight RSNs were extracted: DMN, dorsal attention network (DAN), visual network (VN), language network (LN), cerebellar network (CN), salience network (SN), frontal-parietal network (FPN), and sensorimotor network (SMN).

Our previous findings suggested that the DMN exhibited the most extensive changes in ALFF among RSNs (Shao et al. 2025), which motivated us to prioritize the DMN as the core focus of the current study. This study aimed to further investigate the time-series correlations between the major DMN nodes and whole-brain voxels during the resting state in patients with MCI. Thus, we performed seed-based analyses of SFC across the whole brain. Based on findings from prior studies (Yuan et al. 2021), we selected four key subregions within DMN as ROIs: PCC, MPFC, AG_R and AG_L (Fig. 1). These ROIs were defined as 5-mm spherical regions centered on the peak MNI coordinates of each subregion. The voxel-wise SFC values were obtained, and Fisher-Z transformation was applied to quantify these values, generating a brain map that illustrates the differences in SFC values.

Fig. 1.

Brain regions of interest

Statistical Analysis

SPSS 25.0 software was used to analyze differences in demographic and neurocognitive scale data between the HC and MCI groups. Data were expressed as mean ± standard deviation, with a p-value < 0.05 considered statistically significant. The Shapiro–Wilk test was used to assess the normality of all variables prior to analysis. Continuous data were analyzed using independent t-tests (age, gender, educational level, MoCA, and MMSE scores), while categorical data (gender) were assessed using the Chi-square test.

Two-sample t-tests were conducted to compare the SFC values of different ROIs between the MCI and HC groups using SPM12 within MATLAB R2022b, with age, gender and educational level included as covariates to control for potential confounding on SFC results. For the SFC analysis, statistical significance was defined as a voxel-level threshold of p < 0.001, with cluster-level false discovery rate (FDR) correction at p < 0.05 and a minimum cluster size of > 30 voxels. This cluster size was chosen to balance statistical power and biological relevance, minimizing type I errors. Significant brain clusters and their corresponding coordinates were identified and visualized using xjView and BrainNet Viewer. Correlation analyses were performed to assess the relationships between the identified SFC differences and MoCA scores. Scatter plots for statistically significant results were created using GraphPad Prism.

Results

Demographic and Clinical Characteristics

The demographic and neuropsychological characteristics of the participants are detailed in Table 3. On comparing the MCI group with the HC group, the differences in age, gender, and educational level were insignificant (p > 0.05). However, MoCA and MMSE scores were significantly different between the two groups ( p < 0.001).

Table 3.

Demographic and clinical data of the subjects

Clinical variables	MCI(n = 36)	HC(n = 26)	χ2/t	p-value
Age (years)	67.75 ± 6.88	64.08 ± 8.04	1.934	0.329
Gender (male/female)	14/22	8/18	0.651	0.518
Educational level (years)	9.94 ± 2.63	10.73 ± 2.84	-1.124	0.265
MoCA scores	17.72 ± 5.16	27.73 ± 1.46	-9.601	< 0.001
MMSE scores	25.92 ± 1.46	28.35 ± 1.09	-7.1148	< 0.001

Altered SFC of DMN in Patients with MCI

Two-sample t-tests were conducted for each of the four DMN seed regions. Compared with HCs, patients with MCI exhibited both increased and decreased SFC between each seed and multiple brain regions. In brief, the PCC and MPFC seeds showed a mix of increased and decreased SFC, while the AG_R and AG_L displayed primarily decreased SFC. These findings suggest that the DMN-related functional networks may undergo varying degrees of reorganization in MCI.

The results with the PCC as the seed region showed that SFC values between the PCC and Insula_R and Cerebelum_Crus1_R were increased, whereas SFC values between the PCC and each of Frontal_Med_Orb_R, Fusiform_L, Temporal_Pole_Mid_R, Temporal_Pole_Sup_R, and Rectus_R were decreased (see Fig. 2a).

Fig. 2.

Brain regions with significant differences in SFC values in patients with MCI. (Warm colors represent regions of significantly increased SFC values in the MCI group compared with those in the HC group; cool colors represent regions of decreased SFC values) (a)PCC as the seed point; (b) MPFC as the seed point; (c) AG_R as the seed point; (d) AG_L as the seed point

The results based on the MPFC seed revealed that the MCI group showed increased SFC with the Frontal_Inf_Oper_R, Frontal_Sup_L, Frontal_Sup_R and Caudate_R. In contrast, decreased connectivity was observed with the Lingual_R, Cerebelum_9_L, Calcarine_R, Postcentral_L, Cerebelum_6_L, and Fusiform_L (see Fig. 2b).

In the MCI group, SFC values between AG_R and the Temporal_Pole_Sup_L and Insula_R were increased. Meanwhile SFC values decreased between AG_R and Caudate_L, Rectus_R, Cingulum_Mid_R, Calcarine_R, and Cuneus_R (Fig. 2c).

The MCI group exhibited increased SFC values between AG_L and the Caudate_R but decreased values between AG_L and Fusiform_L (p < 0.001, voxel > 30; Figs. 2d and 3).

Fig. 3.

A 3D Brain network map of SFC changes in the MCI group compared to those in the HC group. Green, ROIs; red, positive clusters; blue, negative clusters. (a) PCC; (b) MPFC; (c) AG_R; (d) AG_L

Correlation Analysis

We analyzed the correlation between DMN SFC and cognitive behavioral index (MoCA score). Brain regions exhibiting significant between-group SFC differences were defined as novel ROIs. For both MCI and HC groups, mean SFC values within these novel ROIs were extracted to represent regional functional connectivity strength; Pearson correlation analysis was selected to examine the relationship between ROI SFC values and the MoCA score based on the normality test results (p < 0.05). Cluster-level FDR correction (p < 0.05) was applied for multiple comparisons. No significant correlation was found between SFC values of these ROIs and MoCA scale in the HC group. In the MCI group, a significant negative correlation was observed between SFC values of the Insula_R and MoCA scale (r = -0.385, p < 0.05, FDR-corrected). Conversely, no significant correlation was found between the MoCA scale and other brain regions (Fig. 4).

Fig. 4.

(a) Correlations between SFC differences and neuropsychological scores. The abscissa is the MoCA score, and the ordinate is the SFC difference intensity of the insula; (b) Three views of the cluster of insula R (sagittal, coronal, and axial)

Discussion

Previous studies have predominantly focused on SFC within the DMN in patients with MCI. This study is the first to investigate whole-brain SFC alterations using four key DMN subregions as seed points (PCC, MPFC, AG_R and AG_L). Main findings included: (1) The DMN exhibited both increased and decreased SFC with multiple brain regions outside and within the network, with decreased FC being more widespread than increased FC. The regions with reduced FC were primarily located in the precuneus, frontal gyri, temporal gyri, insula, caudate nucleus, lingual gyrus, fusiform gyrus, and postcentral gyrus. (2) The altered SFC of DMN subregions was linked to episodic memory disturbance in MCI. (3) AG_R-insula SFC was negatively correlated with MoCA scores.(4) It reflects abnormal interaction patterns between the DMN and other brain networks.

The human brain is a complex network comprising several relatively independent but interconnected regions. Of these, the DMN is a core network that maintains spontaneous activity during resting states and primarily contributes to memory integration, attention distribution, self-referential processing, and cognition (Eyler et al. 2019). Episodic memory impairment, a hallmark feature of AD, is also the most common cognitive deficit in MCI (Knopman and Petersen 2014). Previous studies have revealed abnormal activity in the PCC, MPFC, and AG during episodic memory encoding and retrieval (Wang et al. 2016), and that impaired episodic memory is associated with reduced SFC between the precuneus and other regions. Another study, using coefficient Granger causality analysis with the PCC as the region of interest, found reduced effective connectivity from the PCC to the whole brain, particularly involving the left prefrontal cortex, left medial temporal lobe, and left fusiform gyrus (Yang et al. 2017). Our results, which use four DMN subregions as regions of interest, are similar but not entirely consistent with these findings. Nevertheless, they collectively support the notion that abnormal SFC between DMN subregions and widespread brain areas is causally linked to episodic memory impairment in MCI.

By examining Figs. 2 and 3, the DMN exhibits both increased and decreased SFC with multiple brain regions, with the areas showing decreased SFC being more widely distributed than those with increased SFC. Specifically, the increased SFC values in the MCI group were mainly observed in the precuneus, superior frontal gyrus, inferior frontal gyrus, insula, caudate nucleus, and superior temporal gyrus. However, decreased SFC values were mainly noted in the middle frontal gyrus, superior temporal gyrus, middle temporal gyrus, fusiform gyrus, lingual gyrus, posterior central gyrus, and rectus gyrus. This pattern of whole-brain SFC alterations based on four core nodes of the DMN reflects a decline in functional integration within the DMN and confirms the hypothesis concerning cerebral functional impairment and compensation. An interesting study showed that the MCI group exhibited weakened FC between the cerebellum and the superior and inferior frontal gyri, but strengthened FC between the cerebellum and the middle frontal gyrus (F et al. 2021). In contrast, the present study showed increased connectivity in the superior frontal gyrus and inferior frontal gyrus, but decreased connectivity in the middle frontal gyrus. The discrepancy between the two findings may be attributed to transient neuronal hyperactivation during the early stages of cognitive decline in MCI patients, a phenomenon commonly referred to as “hyperconnectivity.” (Hillary and Grafman 2017) This also explains the differences in AG_L vs. AG_R connectivity patterns: AG_R is characterized primarily by compensatory hyperconnectivity, whereas AG_L is dominated by marked decreases in functional connectivity. Such compensatory activity may also be observed in the lingual gyrus and insula (Knopman and Petersen 2014, Wang et al. 2016). Previous studies have confirmed gradual and qualitative neurofunctional differences between the MCI subgroups, suggesting the existence of compensatory and deterioration mechanisms in functional networks, which are primarily derived from the DMN (Varela-López et al. 2025).

Notably, the negative correlation between AG_R-insula SFC and MoCA score indicates that enhanced functional connectivity in this circuit is associated with poorer cognitive performance in MCI, which is consistent with the compensation hypothesis. For the first time, we found a direct association between the connectivity of this circuit and cognitive function. Therefore, we speculate that the fewer brain regions with enhanced SFC values, the greater the neuronal damage and the higher the risk of MCI progressing to AD. As a supplementary note, compensatory overactivation has been hypothesized to exhibit a positive association with cognitive function by certain scholars, which contrasts with our experimental results (Guan et al. 2022, Simon et al. 2020). We also found that no significant correlations were observed in the other three DMN seed points, which may be related to the stage-dependent progression of compensatory and degenerative processes (Varela-López et al. 2025).

Patients with MCI often present with impairment in the integrity of multiple brain networks (Chhatwal et al. 2018), with extensive differences in FC among multiple cortical networks (Jones et al. 2016), such as the insula (in the SN), frontal lobe and caudate nucleus (in the FPN), and the postcentral gyrus of the DAN, which is consistent with previous studies. The DMN, SN, and CEN, key components of the triple network, have become the focus of recent research. DMN is active at rest and deactivates during complex tasks, ECN is involved in cognitive, decision-making and executive function, while SN maintains attention and regulates emotion. Growing evidence suggests that the “triple network model” closely interacts to regulate higher-order cognitive functions and is differentially impaired across the AD spectrum (Li et al. 2019, Feng et al. 2023, Zheng et al. 2024). In contrast to previous studies, our findings demonstrate SFC abnormalities among subregions of the DMN, SN, and CEN in MCI, reveal widespread FC reductions in cortical and subcortical regions (e.g., insula, caudate nucleus, postcentral gyrus), and identify specific subregions (e.g., SN insula, CEN frontal areas) closely associated with episodic memory impairment. More studies have shown that the triple network can be used to detect the reliability and stability of large-scale connections, which is closely related to cognitive impairment in patients with AD (Kemik et al. 2024). Numerous studies have found that FC values among multiple brain networks, such as the DMN, SN, and CEN, in patients with AD and MCI significantly altered (Feng et al. 2023, Zheng et al. 2024, Xue et al. 2021).(Soman et al 2020). also found a reduction in FC across networks such as the DMN, DAN, SMN, and FPN in patients with AD and MCI (Soman et al.2020). These findings extend well beyond the DMN, reasonably suggesting that the functional alterations result from the combined effects of multiple brain regions or even multiple brain network systems. This further indicates that MCI is associated with abnormal interaction patterns across large-scale functional brain networks.

Limitations

This study has several limitations. A total of 41 patients with MCI and 33 HCs were included in the experiment; the relatively smaller sample size may limit the generalizability of our findings. To address this limitation, we are actively recruiting additional participants. Furthermore, the potential confounding effects of age, sex, and educational level distribution could influence the study’s outcomes; however, we accounted for these variables as covariates in the statistical analyses, thus minimizing bias. Additionally, our cross-sectional design only provides baseline data, which limits causal inference regarding FC changes and episodic memory—core cognitive functions impaired in AD. Future research should consider longitudinal assessments of FC dynamics across the AD spectrum. Moreover, integrating more emerging sequences and cutting-edge analytical techniques and algorithms would allow for a more comprehensive evaluation of neuroimaging markers throughout AD progression, expanding the relevance of this study.

Conclusion

In conclusion, this study mainly reveals significant changes in SFC between DMN subregions and other cognitive-related brain regions in patients with MCI compared with HCs. The results showed that regions with reduced SFC values were more widely distributed than those with enhanced SFC values. These SFC alterations may be associated with episodic memory impairment, suggesting that MCI may reflect early functional changes and compensatory mechanisms preceding structural damage. This pattern of whole-brain SFC alterations based on the core nodes of the DMN also reflects an abnormal reorganization between the DMN and other brain networks. The identification of sensitive neuroimaging biomarkers in the SFC profiles of the DMN may provide new insights for early detection of AD. Notably, these findings could guide the development of targeted neuroimaging tools for the early screening of AD, further highlighting the clinical relevance of our work.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Y.Z. contributed to the experimental design of the study, the analysis and interpretation of the data, and manuscript writing. J.S. contributed to the analysis and interpretation of the data, as well as manuscript writing and revision. D.Z. performed and analyzed the experiments. M.Q. provided financial support. Z.W. was involved in data analysis. Q.X. was involved in the experimental design and overall supervision, manuscript revision, and financial and grant support. All authors have approved the final version of the manuscript for publication.

Funding

This study has received funding by the Science and Technology Commission of Shanghai Municipality (Grant No. 23Y11907500), the Outstanding Leaders Training Program of Pudong Health Bureau of Shanghai (Grant No. PWRl2022-05), and the Commission of Shanghai Municipal Health (Grant No. 20254Y0168).

Data Availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.

Consent for Publication

Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.

Ethics Approval and Consent to Participate

The studies involving human participants were reviewed and approved by the Ethics Review Committee of Shanghai East Hospital (approval number: No. 2022 − 206; approval date: July 15, 2022). The patients/participants provided their written informed consent to participate in this study.