Introduction

With advancing age, crystallized cognitive abilities, such as accumulated knowledge and vocabulary, tend to be preserved or even improve, while fluid cognitive abilities, including abstract thinking, reasoning, and decision-making, decline in older adults, independent of pathology [1–3]. These declines in fluid abilities are associated with age-related deficits in processing speed, attention, memory, and executive function, collectively known as cognitive aging [1, 2]. Cognitive aging is linked to functional impairments in managing new information, adapting to the environment, and problem-solving, impacting older adults’ independence in society and at home [4, 5]. Extensive evidence suggests that cognitive decline in older adults is associated with changes in brain structure and function, emphasizing the importance of identifying brain-based factors contributing to cognitive aging [3, 6–8]. Understanding age-related neural alterations in cognitive aging may aid in differentiating normal aging from disease states and identifying brain areas for intervention [3].

Concerning individuals, aging correlates with cognitive decline, even among those who maintain normal daily activities [9]. Research indicates widespread changes in the brain, including reductions in grey matter, disruptions in anatomical connections, and diminished functional connectivity (FC). These findings underscore the complexity of cognitive aging and emphasize the necessity of exploring its neural underpinnings for better understanding and potential intervention [9].

The natural decline in cognitive abilities during healthy aging, even without pathology, significantly impacts daily activities, quality of life, and has psychological and socio-economic consequences. Differentiating between normal cognitive decline and early signs of dementia or mild cognitive impairment (MCI) remains challenging, making early detection crucial for effective intervention [10].

Though the numerous studies have established that functional networks in the aging brain differ from those in younger individuals [10], the exact nature and extent of these changes remain incompletely understood. They are believed to result from a combination of cell loss, functional decline, and emerging compensatory mechanisms. Additionally, associations have been found between these altered resting-state FC patterns and cognitive performance in older adults [3, 11]. Moreover, FC disruptions are evident in clinical conditions prevalent among the elderly, such as Alzheimer’s disease (AD), Parkinson’s disease, and MCI, with the severity of symptoms correlating with connectivity metrics [10, 12].

Even though “age” significantly impacts functional and anatomical brain connectivity, more research is needed to understand these changes fully due to inconsistent findings. While most studies report decreased FC, some find increased between-network connectivity. Onoda et al. [11] and Huang et al. [13] found no age-related association. These inconsistencies may be due to differences in study designs and methodologies, highlighting the need for further investigation [9].

The United Nations World Population Prospects predict that the number of people over 80 will triple by 2050, with the group 65 or older is experiencing the fastest growth rates worldwide. However, significant regional disparities exist in these demographic trends [14]. These demographic shifts pose challenges for healthcare management of age-associated diseases like cancer, neurodegenerative disorders, and dementia. Addressing these challenges requires a multidisciplinary approach spanning various scientific fields [15].

Recent advances in techniques such as magnetic resonance imaging (MRI), magnetoencephalography (MEG), and electroencephalography (EEG) have significantly enhanced our ability to analyze brain networks, providing deeper insights into the complex relationships between aging and dementia. These neuroimaging tools have made it easier to study changes in FC that are linked to cognitive decline. This has led to a better understanding of how brain networks are affected by both normal aging and neurodegenerative diseases. In this review, we aim to summarize recent findings derived from these three techniques, highlighting their contributions to the ongoing exploration of FC in aging and dementia.

FC and cognitive decline

Understanding and introducing MCI and FC is crucial in the study of cognitive decline and neurodegenerative diseases. MCI serves as an intermediate stage between normal cognitive aging and conditions like AD, making it a key focus for early detection and intervention strategies. Recognizing the patterns of cognitive decline in MCI allows researchers to differentiate between normal aging and pathological processes, aiding in timely diagnosis.

FC, which refers to the coordinated activity between different brain regions, offers valuable insights into how brain networks change with age. Both cognitive performance and the progression of neurodegenerative diseases are associated with alteration in FC. By studying FC in individuals with MCI, leading to understand the neural mechanisms underlying cognitive decline with more effective treatments and interventions aimed at slowing disease progression.

MCI is characterized by self- and hetero-reported cognitive complaints, objective cognitive impairment, and preserved independence in daily activities without meeting the criteria for dementia [16]. MCI often represents a transitional state from healthy aging to AD or other forms of clinical dementia. However, it is crucial to recognize that not all MCI cases progress to dementia [15].

In recent research, there has been a growing focus on understanding the functional and structural impacts of MCI in comparison to healthy aging and AD. Studies utilizing blood oxygen level dependent (BOLD) contrast have extensively analyzed differences in FC or coactivation patterns between brain areas in both task-based and resting-state designs [15].

People who have MCI often have cortical atrophy, less activity in certain part of the brain, and more activity in the default mode network (DMN) when they are at rest [17]. Furthermore, researchers observe a more severe decrease in FC in MCI patients who eventually transition to AD [18]. For example, Sullivan et al. [19] found that cognitive preservation in healthy older adults is associated with higher interactivity across brain regions, while MCI participants show lower interactivity, which correlates with lower Mini Mental State Examination scores. According to Zhao et al. [20], MCI patients have lower global and local efficiency in whole-brain FC networks compare to healthy controls [15].

Recent studies have also highlighted the heterogeneity within MCI subtypes and their distinct implications for FC changes. For instance, research has identified different MCI subtypes based on patterns of cognitive decline and progression risk, such as low-, medium-, and high-risk subtypes. Each subtype demonstrates unique FC profiles, which may have significant implications for understanding disease mechanisms and tailoring interventions. High-risk MCI subtypes often exhibit more pronounced disruptions in FC and greater cortical atrophy compared to low-risk subtypes. This variability underscores the importance of differentiating between MCI subtypes to better understand their specific neural signatures and progression trajectories.

Addressing these findings, it is evident that MCI is not a homogeneous condition and that FC changes may vary significantly among subtypes. This heterogeneity in MCI subtypes emphasizes the need for more refined and individualized approaches in both research and clinical practice to enhance early diagnosis and treatment strategies for dementia [21].

In addition, it is mentioned that the two highly prevalent subtypes of MCI are amnestic MCI and vascular MCI. Both subtypes exhibit widespread structural and functional brain alterations, yet the patterns of FC strength (FCS) alterations remain inadequately explored. A study examining resting-state FC changes among individuals with amnestic MCI, amyloid beta, and healthy controls found significant reductions in FCS in the left middle temporal gyrus for both groups compared to healthy controls. The study revealed a significant reduction in FC between the left middle temporal gyrus and various frontal brain regions, particularly in participants with vascular MCI. Furthermore, a partial correlation analysis revealed a positive correlation between the FC values of the left middle temporal gyrus and the left inferior frontal gyrus with episodic memory performance, and a negative correlation with living status. These findings provide new insights into the pathophysiological mechanisms underlying different MCI subtypes [22].

FC of the brain network, defined as the temporal correlation between activity measures from distributed brain regions, is analyzed using various metrics focusing on synchronization of different brain rhythms. These networks’ topology changes with human activity, offering insights into cognitive load, working memory, and skill acquisition. Alterations in motor cortex connectivity reflect motor learning and are vital for brain computer interface development. FC serves as a diagnostic tool for neurological disorders like autism spectrum disorder, AD, depression, and multiple sclerosis, indicating atypical FC networks. Moreover, FC networks are sensitive to age-related brain changes and provide valuable insights into physical and cognitive health in late life [9].

FC between distinct brain regions offers important information on how different brain regions cooperate to support different types of behaviors and cognitive tasks. As mentioned above, FC represents the temporal correlations in neural activity among different brain regions, suggesting that these regions are interacting or collaborating to process information. The coordination of various brain regions for tasks like problem-solving and decision-making is known as cognitive integration, and it depends on this communication. Additionally, it makes sure that various cognitive processes—like memory and attention—align and coordinate when doing challenging tasks [23].

Large-scale networks in the brain interact to support various functions. Key networks involved in introspective thinking, cognitive control, and relevance detection are the DMN, executive control network (ECN), and salience network (SN), in order. FC between these networks is vital for balancing between internal thoughts and external stimuli. Disruptions in FC within and across these networks can cause cognitive deficits and are common in neurodegenerative diseases. This show how important these connections are for keeping your brain healthy.

FC also provides insight on the relative contributions of certain brain regions to particular cognitive functions. For example, connections between the prefrontal cortex and parietal regions promote attention and working memory activities, whereas coordination between the hippocampus and cortical areas is essential for memory encoding and retrieval. The limbic system and prefrontal cortex’s connectivity influence emotional regulation. Therefore, the FC patterns show how different brain regions work together to promote emotional and cognitive tasks [23].

Variations in FC can indicate pathogenic conditions as well as adaptive mechanisms. In aging or neurodegenerative diseases, the brain may enhance connectivity in certain regions to compensate for losses in others. A number of neurological disorders, including AD, are frequently characterized by abnormal FC patterns. The altered connection between the DMN, ECN, and SN in AD reflects both the disease’s development and cognitive loss. These alterations may function as biomarkers for methods of early detection and intervention [23].

On a more granular level, FC relates to synaptic and cellular mechanisms. It reflects the effectiveness of synaptic connection, which is essential for efficient neuronal transmission between various brain regions. Disruptions in this connection frequently correlated with cognitive declines and illness states. Further evidence that changes in neurotransmission may impact cognitive performance comes from FC’s connection to neurotransmitter systems that regulate brain activity [23, 24].

Understanding the biological implications of FC helps explain how neural networks integrate to support cognitive functions and how disruptions in these connections can signal normal developmental changes or pathological conditions. This knowledge is crucial for developing targeted interventions to address cognitive decline and neurodegenerative diseases, as it sheds light on the underlying mechanisms contributing to these conditions [23, 24].

Recent findings of resting-state functional MRI (rs-fMRI)

rs-fMRI has become increasingly popular for studying age and pathology-related changes in the brain. It detects coherent fluctuations in brain activity when individuals are not actively engaging in cognitive tasks [3].

rs-fMRI offers significant advantages for studying aging populations. It enables the examination of FC without the need for a task, mitigating potential confounds from cognitive or motor impairments. Moreover, the relatively short scan times (5–15 min) are beneficial for older adults who may struggle with prolonged periods of lying on their back or have psychological concerns about the scanning environment [3].

rs-fMRI has revealed low-frequency resting-state networks across the lifespan, delineating the brain’s functional architecture. These networks cover attention, memory, cognitive control, default mode, motor, and sensory systems [36]. They were further derived by Yeo et al. [37] further categorized them into seven major resting-state networks. These are the DMN, DAN, frontoparietal control network (FPCN), cingulo-opercular network (CON), limbic network, visual network, and somatomotor network [3].

Investigators have used rs-fMRI techniques to examine how the above patterns of resting-state network FC change with age [8]. While it is evident that healthy older adults disrupt resting-state network connectivity, the contribution of these age-related changes to the cognitive aging process remains unclear. Researchers looking into the link between network connectivity and cognition in older adults have used a variety of cognitive tasks and resting-state methods. The results have been inconsistent, suggesting that future research should include all networks in analyses instead of just a few and incorporate sensitive cognitive measures that reflect the cognitive aging process [3].

One recent study has been conducted to identify a relationship between resting-state network connectivity and the overall cognitive aging process (Table 1). In this study, the NIH Toolbox Cognition battery [38] and the rs-fMRI data were used to look at how the four higher-order cognitive networks (DMN, DAN, FPCN, CON) affected the fluid cognition composite scores in a large sample of healthy older adults. This study provides important insights into inter-individual differences in resting-state network connectivity and the cognitive aging process. Specifically, it identifies CON coherence as a potential marker for fluid cognitive performance in nonpathological aging. Among the four higher-order cognitive networks mentioned above, CON connectivity showed the strongest relationship with general fluid cognitive ability in a large sample of healthy older adults. CON connectivity explained more variance in fluid cognitive performance than age and education, affecting three out of five fluid cognitive subtests, indicating a global influence on cognitive aging. These findings suggest that CON connectivity is a key factor in cognitive aging and should be a focus in future research on neural substrates of age-related cognitive decline.

The study has performed by aiming to examine the effects of age and cognitive effort on resting-state FC (RSFC) of the hippocampus (HC) in older adults. It involved fMRI scans and seed-based RSFC analysis. The researchers also aimed to examine the correlations between RSFC before and after cognitive effort for distinct HC connections and whether these associations changed with age. The study also aimed to determine if changes in intra-HC RSFC following effort may predict future memory decline in older adults [39].

The findings indicated that for younger persons, increases in RSFC between the DMN hubs and the left anterior hippocampus (laHC) were associated with post-effort increases in laHC-pHC RSFC. On the other hand, in older people, reductions in the RSFC of the right precentral gyrus and laHC were associated with post-effort increases in the RSFC of laHC-pHC. Thus, age and cognitive exertion changed the connection between intra-HC and inter-HC RSFC. Crucially, two years later, older persons with decreased post-effort RSFC between the laHC and the posterior hippocampi (pHC) showed a deterioration in episodic memory. Therefore, according to this study, the alteration in intra-HC RSFC after cognitive activity was able to predict memory function later in life [39].

A recent study by Liu et al. [40] in radiology compared fMRI scans in 614 healthy individuals aged 18–88 years to understand how brain function evolves over the human lifespan. This research contributes to the clinical practice of radiologists by examining the inner workings of the brain [41].

Studying brain network coactivation patterns (CAPs) can help explain how, when, and where brain networks coactivate. Liu et al. [40] identified six CAPs between brain networks, with three correlated with age. They calculated dwell time and transition time, finding that one dwell time and three transitions correlated with age. This study helps understand how brain networks coactivate and correlate with each other. The authors found a reduction in interactions between the frontoparietal network and the DPN with older age, possibly due to lower cognitive flexibility. Primary sensory functions remain intact. They also found evidence of cortical plasticity in the brains of older people. The DPN and attentional network CAPs became dominance with age, making up for the primary sensory network and frontoparietal network CAP becoming weaker [41].

Deery et al. [42] have summarized that the brain undergoes significant structural, functional, and metabolic changes with age, leading to alterations in cognition and behavior. In the early years, functional networks are rapidly organized and refined until around the third and fourth decade. These changes can lead to various expressions of cognition and behavior. Older adults display lower within-network connectivity and higher between-network connectivity than younger adults. The trajectory of alterations varies by network. Some networks may remain relatively stable (e.g., primary sensory and motor) and others may vary substantially (e.g., associative regions) [42].

The typology of functional networks influences age-related changes and their functions. Older adults have a less segregated, modular, and integrated system of networks than younger adults. They have lower efficiency in higher-order networks. Alterations to a “rich club” network hub reduce communication between brain regions, impacting efficient higher-order cognitive functions. These functional network changes are likely driving aging in higher-order cognitive processes [42].

Functional networks based on glucose metabolism show covariance differences across adult life, but evidence is limited. The “small world” properties of metabolic brain networks show strong correlation between brain glucose metabolism and local and global connectivity, especially for hubs with high between-network connectivity. This suggests brain glucose is a key contributor to age-related changes in connectivity and cognitive performance [42].

Recent MR-PET measurements show synchrony between hemodynamic processes and glucose uptake, offering high sensitivity and regional specificity for functional and multimodal brain imaging. However, their application in aging and cognitive decline has not been widespread. In cognitive aging, there is a well-documented complex interaction between brain glucose, age, and peripheral physiology, including cardiovascular and metabolic factors. Future research will need to characterize these peripheral factors [42].

Recent findings of EEG and MEG

EEG and MEG are non-invasive neurophysiological techniques that record electromagnetic signals from the ionic currents created by cerebral activity at the microscale. Comparing EEG to other neuroimaging technologies, it is also more affordable and adaptable. With a high temporal resolution (a few milliseconds), neurophysiological techniques allow the study of brain rhythms produced by the rhythmic activity of huge groups of cortical neurons (mesoscale) [51]. Five primary frequency bands allow brain areas to communicate: beta (12–30 Hz), alpha (8–12 Hz), theta (4–8 Hz), delta (0.5–4 Hz), and gamma (35–100 Hz). Electrophysiological signals reflect the functional synchronization (or desynchronization) produced by the oscillation of large groups of cortical neurons of several neural systems (macro-scale) supporting alertness, motivation, and several cognitive processes [52].

The interactions between inputs and outputs of neural systems can be represented by various dimensions, including deterministic-stochastic, complexity-simplicity, linear nonlinear, stationary-nonstationary, and phase-non phase locking [51].

There are several methods to evaluate FC between brain areas based on signal synchronization and coupling (Table 2) [52].

These measures can detect neuropathophysiological processes before clinical symptoms and structural alterations, proving valuable for clinical applications and new theoretical considerations [51]. Studies have shown high accuracy in distinguishing MCI from healthy controls using machine learning (ML) and synchrony measures at resting state, sensitive to early AD changes [53]. For example, Dauwels and Vialatte [54] achieved 80–85% accuracy in classifying HC and MCI using resting-state EEG (rsEEG) measures. Musaeus et al. [55] achieved 95% diagnostic accuracy in distinguishing HC, MCI, and AD patients using coherence, imaginary part of coherency (iCOh), and weighted phase lag index (PLI), noting decreased alpha coherence in AD patients. These findings highlight the potential of these methods in early AD diagnosis [52].

However, findings on FC changes in MCI are inconsistent. Studies have reported both global hyper synchronization and hypo synchronization in various EEG bands compared to healthy controls. Many studies observed a slowdown of EEG activity in MCI, with a significant reduction in alpha activity at rest noted as a potential AD biomarker. Local activity alone might not fully explain the temporal dynamics of neurodegenerative diseases as “disconnection syndromes”. Understanding this intermediate stage is crucial for finding neurofunctional biomarkers, as the boundaries between aging and MCI blur. A systematic review of brain connectivity alterations in MCI is necessary [52].

A recently conducted meta-analysis [52] aimed to identify early signs of connectivity changes in MCI by quantitatively analyzing previous findings on synchrony alterations in MCI patients compared to healthy controls at resting state. They have calculated the standardized mean difference (SMD) of effect sizes from studies found through systematic literature research up to June 2023. Additionally, they have employed a novel random-effects analysis method, “MetaNSUE”, which includes non-statistically significant unreported effect sizes. This is the first study to use this innovative approach in a meta-analysis of synchrony changes in pathological aging. Identifying an electrophysiological biomarker of AD could facilitate early and accessible diagnosis, improving disease management. The study reveals a global reduction in functional integration between cortical areas, indicating a “disconnection syndrome” that could be observed even at the MCI stage in specific brain sites. The electrophysiological assessment of temporo-parietal functional synchronization in the alpha band is used to detect neurodegenerative processes, such as gray matter atrophy and temporo-parietal hypometabolism, which are well-established neuropathological biomarkers of AD. ML studies show that alpha 1 sources in parietal and temporal areas are stronger in MCI converters than stable subjects.

Several previous studies of resting and task-related EEG coherence in AD patients have demonstrated reduced synchronization within and between hemispheres. A recent study group hypothesized that if synaptic disconnection as the neuropathology of AD is responsible for the failure of the brain to integrate various regions into effective networks, then electroencephalographic evidence of the disruption of FC might be used to diagnose AD. They have explored the relationship between EEG coherence and executive function in patients with MCI, AD, and healthy controls. The study’s general goal was to identify potential AD markers that could facilitate early recognition of neural functional deficiency and reduce the time needed to diagnose major neurocognitive disorders in primary care [56].

The above study found reduced FC in all studied networks in AD patients, with frontotemporal and parietal beta wave coherence significantly lower in the AD group. The study suggests that task-related EEG coherence in the AD group was modulated by cognitive performance demands. The most notable changes were found in frontal EEG coherence during the visual-spatial task challenge, parietal EEG coherence during the writing task, and temporal EEG coherence during the speech understanding task. The results suggest that changes in spontaneous EEG coherence of the frontoparietal region in the AD group can be detected in a clinical setting, potentially serving as a marker of neuronal disconnection. The MCI group demonstrated a paradoxical increase in task-related EEG coherence in temporal lobes, suggesting high neurocognitive reserve and increased compensatory activities or an alternative to AD neurodegenerative processes. The sensitivity of EEG coherence as a marker of neuronal disconnection needs to be explored, and the specificity of EEG coherence markers is another challenge to be addressed [56].

A study examined group differences in EEG coherence within global cortical networks at rest and during executive challenges among patients with AD, individuals with MCI, and healthy controls. Results showed decreased EEG coherence in cross-hemisphere frontal, temporal, parietal, and occipital pairs in the AD group both at rest and during tasks requiring comprehension, analysis, perceptual-motor response, and executive functioning. Overall, the results from our study support the disconnection hypothesis of AD, which proposes that cognitive deficits may be due to the diffuse disconnection process in neurocognitive dementias rather than isolated changes in specific areas. The difference in EEG coherence between healthy and AD patients could play an important role in clinical practice. As neurodegeneration starts long before clinical manifestations of AD, detecting neuronal disconnection across hemispheres with EEG might be possible even in the pre-clinical stage. Further evaluation of the markers’ sensitivity and specificity to the neurodegenerative process in the brain needs to be conducted [57].

Another recent study group has mentioned that understanding how MCI affects global neural networks may explain changes in brain electrophysiology. By using graph theory (GT) and the visual oddball paradigm, they have evaluated the FC of neuronal networks in brain lobes. The study has involved 30 participants: 14 with MCI and 16 healthy control participants. They have conducted an examination using the visual oddball paradigm, focusing on EEG signals with targeted stimuli. Their analysis employed FC utilizing the change point detection method. Additionally, they have implemented training for linear discriminant analysis, K-nearest neighbor, and decision tree techniques to classify brain activity, distinguishing between subjects with MCI and those in the healthy control group. The results demonstrate the efficacy of combining FC measurements derived from EEG with ML for cognitive impairment classification. This research opens avenues for further exploration, including the potential for real-time detection of cognitive decline in complex real-world scenarios [58].

Quantitative EEG (QEEG) is useful for predicting treatment responses, but no study has looked at changes in FC using QEEG after a lifestyle intervention. A research group investigated neurophysiological changes in QEEG after a 24-week multidomain lifestyle intervention in the South Korean SUPERBRAIN study. The study involved participants aged 60–79 without dementia but with at least one dementia risk factor. They were randomly assigned to a facility-based intervention (FMI) (n = 51), a home-based intervention (HMI) (n = 51), or a control group (n = 50). Data from 44 FMI, 49 HMI, and 34 control participants who underwent EEG at baseline and study end were analyzed [59].

The study found that participants who received the intervention showed increases in the power of the beta1 and beta3 bands and the iCoh of the alpha1 band compared to the control group. They also showed decreases in the characteristic path lengths of the alpha1 band in specific brain regions, indicating improved functional brain networks and higher global efficiency. This is the first study to demonstrate positive functional brain changes using QEEG after a 24-week lifestyle intervention to prevent dementia. The results suggest increased FC and higher global efficiency in the intervention group. More studies with larger sample sizes and longer intervention periods are needed to confirm these findings. Quantitative EEG helps understand brain health in aging people. Studies on EEG coherence and power in specific brain areas have had mixed results. Another study has measured EEG coherence and power across the entire scalp and a wide frequency range to find global EEG markers for cognition in people at risk for dementia. Global markers are more reliable and less error-prone than specific ones, which could help resolve past inconsistencies. The study has looked at global EEG coherence (1–30 Hz) and an EEG slowing score, which is low-frequency power (1–8 Hz) divided by high-frequency power (9–30 Hz). It also has examined how these EEG measures predicted cognition and cognitive decline over 5 years [59].

The results have shown that baseline global coherence was linked to better cognition at the start but not to cognitive decline or cognition after 5 years. There was no correlation between the EEG slowing score and either cognitive decline or cognition. This suggests that resting state global EEG coherence might be a useful marker for brain health in older adults at risk for dementia. However, the small sample size of study hindered its ability to determine the predictive value of global coherence, a topic that requires further exploration in larger studies [59].

Recent findings of GT approaches

GT, a mathematical tool for quantifying networks, has been used to study cognitive systems where nodes represent tasks and edges represent correlations between performances. Research by Garcia-Ramos et al. [66] and Kellermann et al. [67] on epilepsy demonstrates its utility in identifying neuropsychological differences [68]. Tosi et al. [69] found qualitative differences in cognitive networks between healthy individuals and those with neurological diseases but lacked quantitative differentiation. Ferguson [70] confirmed network reorganization in early Alzheimer’s stages by comparing the cognitive networks of healthy older adults with early Alzheimer’s type dementia and amnestic MCI. However, research on cognitive network changes during healthy aging is limited, which is crucial for detecting topological deviations indicating neuropathology. Different cognitive impairments likely show unique network topologies, as seen in AD versus vascular encephalopathy [69].

Wright et al. [68] have applied GT to compare cognitive network structures among healthy individuals of different ages and patients with varying cognitive impairments. They used six network metrics to assess integration, segregation, and modularity. They then tested to see if differences in neural networks in AD show up similarly in cognition. They were expecting distinct network topologies between healthy aging and pathological cognitive decline. The study hypothesized that older, healthy adults rely more on semantic memory within their cognitive networks, whereas individuals with AD, even at prodromal stages, would show reduced influence of semantic processing. For Alzheimer’s patients, network differences have been anticipated in semantic processing and memory functions. In healthy aging, older adults’ networks have expected to emphasize crystallized abilities with notable differences in executive function tests due to age-related declines. Additionally, individuals with amnestic MCI have expected to have network compositions similar to those with Alzheimer’s dementia, aligning with Ferguson [70] and other neuroimaging studies [68].

Finally, the study demonstrated that cognitive networks change with aging and are differentially affected by cognitive impairments, with specific alterations potentially serving as early indicators of AD. These findings could help develop better diagnostic tools for neurodegenerative diseases [68].

According to Bateman et al. [71], AD has a long preclinical phase where brain changes occur without obvious cognitive symptoms. Changes begin 25 years before symptoms show up in familial AD. They start with less amyloid in the cerebrospinal fluid, then amyloid deposition, increase tau protein, brain atrophy, and hypometabolism [72]. Vermunt et al. [73] have studied a large cohort from the Dominantly Inherited Alzheimer Network (DIAN) using structural MRI and GT. They have found that brain connectivity changes begin as early as 13 years before symptoms, particularly in the precuneus. The correlation between these changes and biomarkers such as amyloid accumulation, brain metabolism, cortical thickness, and cognition suggest the potential use of brain connectivity measures for early diagnosis and tracking AD progression [72].

Cognitive functions depend on communication between interconnected brain regions, not isolated areas. Recent brain network studies use GT, representing the brain as nodes connected by edges, to assess its topological architecture. Vermunt et al. [73] have used this approach to study brain network topology in familial AD mutation carriers. They found network changes 13 years before disease onset, showing shorter network paths or more direct links between brain areas. These changes, most prominent in the precuneus—a crucial brain hub and early amyloid accumulation site Palmqvist et al. [74], confirm previous findings in sporadic AD. The shorter paths in the precuneus might be a compensatory mechanism, increasing connections to counteract pathology, similar to changes seen in individuals at higher genetic risk for sporadic AD [72].

The study by Vermunt et al. [73] has found that, in individuals who will develop AD, brain network paths shorten, followed by a loss of small-world organization and reduced local clusters of connections. This indicates that brain networks become more random over time, losing the balance between global and local connectivity. These changes in network topology precede simpler measures such as degree and density, suggesting that complex measures that combine clusters and paths undergo earlier changes and are associate with amyloid deposition, brain hypometabolism, and cognitive deficits. Thus, network topology alterations could be an important AD biomarker, helping to understand cognitive decline mechanisms. It is still unclear what exactly structural correlations in grey matter reflect; they might be linked to physical white matter tracts or FC. Previous studies show overlap between structural correlations and white matter tracts and between clusters from structural and FC. Identifying these network abnormalities may help understand how amyloid and tau pathology spread, causing cognitive decline. They have mentioned that future research should explore the neurobiological factors driving these changes and replicate findings in other samples [72].

A recent study has RSFC in 772 older adults (55–85 years, 421 males) using a graph-theoretical approach. The results show age-related increases in between-network RSFC and decreases in within-network RSFC, indicating a reorganization towards more pronounced functional network integration. This reorganization particularly affects early sensory networks, such as the visual and sensorimotor networks, and correlates with age-related cognitive performance differences. Additionally, lower RSFC within primary processing networks correlates with lower cognitive performance, while increased RSFC between higher-order networks and the sensorimotor network suggests a compensatory mechanism to maintain cognitive functions. However, this increased between-network RSFC often correlates with worse cognitive performance, supporting the dedifferentiation theory, where less specialized networks lead to cognitive decline [75].

The study also found systematic sex-related network differences: females show more segregation in networks like the default mode and ventral attention network, while males show a more integrated network system, particularly in the sensorimotor network. These findings highlight sex-related connectivity differences, possibly influencing sex-related behavioral functioning. The study emphasizes the importance of sex-stratified analyses, as RSFC patterns differ significantly between older males and females, indicating different reorganizational processes with aging [75].

Future directions

Echeverri-Ocampo et al. [58] study has highlighted the complex relationship between MCI and FC, opening new avenues for future research in neurodegenerative disorders and brain function assessment. As they mention, future research should expand beyond MCI to include other neurodegenerative conditions such as AD, Parkinson’s disease, and other forms of dementia. This broader focus can help identify common and unique FC patterns across different disorders, providing deeper insights into their mechanisms. Additionally, integrating other neuroimaging techniques like fMRI and MEG with EEG could offer a more comprehensive view of brain function and connectivity. Future studies should explore the benefits of combining multiple modalities to improve diagnostic accuracy.

Conclusions

The review article summarizes current research on cognitive aging and neurodegenerative disorders through the lenses of FC studies using rs-fMRI, EEG/MEG, and GT approaches. Age-related changes in cognitive abilities, characterized by preserved crystallized skills and declining fluid abilities, underscore the importance of understanding neural alterations in brain structure and function. FC studies, particularly using rs-fMRI, have illuminated disruptions in resting-state networks like the default mode and FPCNs, offering insights into cognitive aging processes and their implications for neurodegenerative diseases such as AD and MCI.

Moreover, advances in rs-fMRI have revealed FC deficits in AD and MCI, particularly in critical networks like the DMN, highlighting its potential as a diagnostic tool. Despite challenges such as NVU and variability in resting-state conditions, rs-fMRI remains integral to studying brain function across the lifespan and in pathological conditions.

EEG and MEG complement rs-fMRI by providing high temporal resolution for studying brain rhythms and FC. These techniques, combined with ML, show promise in enhancing early diagnosis and understanding connectivity changes in neurodegenerative diseases.

GT, another essential tool, quantifies cognitive networks and reveals topological changes associated with AD and MCI. By elucidating network reorganization and connectivity alterations, GT offers insights into early biomarkers and potential diagnostic strategies.

Future directions include integrating multimodal neuroimaging techniques to improve diagnostic precision and therapeutic interventions for neurodegenerative disorders. This review shows how important it is to use methods from different field in neuroscience research to better understand how cognitive networks change during healthy aging and pathological conditions. This is important for finding good solutions to health problems that come with getting older.