Changes In Functional Brain Networks Research Articles

Altered functional connectivity in default mode network maintains attention task performance in school-age children with frontal lobe tumor

This study examines functional brain network changes in children with frontal lobe tumors (FLT). Ten pediatric FLT patients from Beijing Tiantan Hospital and 20 healthy children were compared in terms of cognitive performance and resting-state functional magnetic resonance imaging (rs-fMRI) connectivity. The FLT group showed lower cognitive performance, particularly in visual and working memory domains, but had comparable attention abilities to the healthy controls. There were notable differences in connectivity between the default mode network (DMN) and sensorimotor network (SMN) in both groups. The FLT group also displayed a significant reduction in local efficiency in the left lateral parietal area within the DMN. Importantly, reduced DMN-SMN connections and increased DMN-lateral prefrontal cortex connectivity may facilitate maintaining attention and memory tasks in FLT children. This study sheds light on how the brains of children with FLT adapt, preserving “normal” attention functions despite frontal lobe damage.

Graph theoretical analysis demonstrates associations among cortical network segregation, β‐amyloid status, and reward‐based learning in older age

AbstractBackgroundAs humans age, brain networks supporting cognition show alterations in functional connectivity patterns consistent with network dedifferentiation. Here we examined the extent to which Alzheimer’s disease (AD)‐related β‐amyloid (Aβ) pathology exacerbates functional dedifferentiation in aging and is associated with reduced probabilistic reward‐based learning.Method27 young and 47 cognitively normal older adults underwent 3T functional magnetic resonance imaging (MRI) while performing a novel probabilistic learning task. The task required participants to learn the identity of real‐estate agents as “winning,” “losing,” or “break‐even” based on their success in selling houses for a profit. We used graph theoretical analysis and applied 400 regions of interest to identify age‐related changes in functional brain networks during early and late learning phases in the task. Older adults were characterized as Aβ‐ (n = 34) or Aβ+ (n = 13) using [11C]Pittsburgh Compound‐B positron emission tomography. Analyses focused on the two graph theory measures commonly linked to dedifferentiation (i.e. loss of segregation): clustering coefficient and modularity.ResultsYoung adults showed better learning performance, followed by Aβ‐ older adults and then Aβ+ older adults, but there was no significant age difference between young and Aβ‐ older adults (Fig. 1). Young adults also showed a higher clustering coefficient, which indicates greater network segregation, followed by Aβ‐ older adults and then Aβ+ older adults (Fig. 2A and 2B). Considering the learning phase, we found differences between the young and both Aβ‐ and Aβ+ older adults in the early learning phase (Fig. 2A), and differences between young and only Aβ+ older adults in the late learning phase (Fig. 2B). Analyses of modularity showed similar patterns of results, though the main effect of group was not significant (Fig. 2C and 2D). There was an interaction between network segregation and group to predict learning performance. Only Aβ‐ older adults with higher network segregation exhibited better learning performance (Fig. 3).ConclusionThese findings indicate Aβ is associated with functional network dedifferentiation and poorer learning performance. For Aβ‐ older adults, successful maintenance of brain network segregation may underlie the preservation of “youth‐like” cognitive performance.

EEG functional connectivity after perinatal stroke.

Impaired cognitive functioning after perinatal stroke has been associated with long-term functional brain network changes. We explored brain functional connectivity using a 64-channel resting-state electroencephalogram in 12 participants, aged 5-14 years with a history of unilateral perinatal arterial ischemic or haemorrhagic stroke. A control group of 16 neurologically healthy subjects was also included-each test subject was compared with multiple control subjects, matched by sex and age. Functional connectomes from the alpha frequency band were calculated for each subject and the differences in network graph metrics between the 2 groups were analyzed. Our results suggest that the functional brain networks of children with perinatal stroke show evidence of disruption even years after the insult and that the scale of changes appears to be influenced by the lesion volume. The networks remain more segregated and show a higher synchronization at both whole-brain and intrahemispheric level. Total interhemispheric strength was higher in children with perinatal stroke compared with healthy controls.

Source connectivity patterns in the default mode network differ between elderly golf-novices and non-golfers

Learning to play golf has high demands on attention and therefore may counteract age-related changes of functional brain networks. This cross-sectional study compared source connectivity in the Default Mode Network (DMN) between elderly golf novices and non-golfers. Four-minute resting-state electroencephalography (128 channels) from 22 elderly people (mean age 67 ± 4.3 years, 55% females) were recorded after completing a 22-week golf learning program or after having continued with normal life. Source connectivity was assessed after co-registration of EEG data with native MRI within pre-defined portions of the DMN in the beta band (14–25 Hz). Non-golfers had significantly higher source connectivity values in the anterior DMN compared to non-golfers. Exploratory correlation analyses did not indicate an association to cognitive performance in either group. Inverse correlations between a marker of external attention with source connectivity of the anterior DMN may suggest a trend in the golf group only, but have to be replicated in future studies. Clinical relevance of these findings remains to be elucidated, but the observed difference in the anterior DMN may provide a starting point to further investigate if and how learning golf may have an impact on physiological age-related cognitive changes.

Growing Brains, Nurturing Minds-Neuroscience as an Educational Tool to Support Students' Development as Life-Long Learners.

Compared to other primates, humans are late bloomers, with exceptionally long childhood and adolescence. The extensive developmental period of humans is thought to facilitate the learning processes required for the growth and maturation of the complex human brain. During the first two and a half decades of life, the human brain is a construction site, and learning processes direct its shaping through experience-dependent neuroplasticity. Formal and informal learning, which generates long-term and accessible knowledge, is mediated by neuroplasticity to create adaptive structural and functional changes in brain networks. Since experience-dependent neuroplasticity is at full force during school years, it holds a tremendous educational opportunity. In order to fulfill this developmental and learning potential, educational practices should be human-brain-friendly and "ride" the neuroplasticity wave. Neuroscience can inform educators about the natural learning mechanisms of the brain to support student learning. This review takes a neuroscientific lens to explore central concepts in education (e.g., mindset, motivation, meaning-making, and attention) and suggests two methods of using neuroscience as an educational tool: teaching students about their brain (content level) and considering the neuro-mechanisms of learning in educational design (design level).

Spatio-temporal dynamics of large-scale electrophysiological networks during cognitive action control in healthy controls and Parkinson's disease patients

Among the cognitive symptoms that are associated with Parkinson's disease (PD), alterations in cognitive action control (CAC) are commonly reported in patients. CAC enables the suppression of an automatic action, in favor of a goal-directed one. The implementation of CAC is time-resolved and arguably associated with dynamic changes in functional brain networks. However, the electrophysiological functional networks involved, their dynamic changes, and how these changes are affected by PD, still remain unknown. In this study, to address this gap of knowledge, 10 PD patients and 10 healthy controls (HC) underwent a Simon task while high-density electroencephalography (HD-EEG) was recorded. Source-level dynamic connectivity matrices were estimated using the phase-locking value in the beta (12-25 Hz) and gamma (30-45 Hz) frequency bands. Temporal independent component analyses were used as a dimension reduction tool to isolate the task-related brain network states. Typical microstate metrics were quantified to investigate the presence of these states at the subject-level. Our results first confirmed that PD patients experienced difficulties in inhibiting automatic responses during the task. At the group-level, we found three functional network states in the beta band that involved fronto-temporal, temporo-cingulate and fronto-frontal connections with typical CAC-related prefrontal and cingulate nodes (e.g., inferior frontal cortex). The presence of these networks did not differ between PD patients and HC when analyzing microstates metrics, and no robust correlations with behavior were found. In the gamma band, five networks were found, including one fronto-temporal network that was identical to the one found in the beta band. These networks also included CAC-related nodes previously identified in different neuroimaging modalities. Similarly to the beta networks, no subject-level differences were found between PD patients and HC. Interestingly, in both frequency bands, the dominant network at the subject-level was never the one that was the most durably modulated by the task. Altogether, this study identified the dynamic functional brain networks observed during CAC, but did not highlight PD-related changes in these networks that might explain behavioral changes. Although other new methods might be needed to investigate the presence of task-related networks at the subject-level, this study still highlights that task-based dynamic functional connectivity is a promising approach in understanding the cognitive dysfunctions observed in PD and beyond.

Electroconvulsive Therapy-Induced Changes in Functional Brain Network of Major Depressive Disorder Patients: A Longitudinal Resting-State Electroencephalography Study.

ObjectivesSeveral studies have shown abnormal network topology in patients with major depressive disorder (MDD). However, changes in functional brain networks associated with electroconvulsive therapy (ECT) remission based on electroencephalography (EEG) signals have yet to be investigated.MethodsNineteen-channel resting-state eyes-closed EEG signals were collected from 24 MDD patients pre- and post-ECT treatment. Functional brain networks were constructed by using various coupling methods and binarization techniques. Changes in functional connectivity and network metrics after ECT treatment and relationships between network metrics and clinical symptoms were explored.ResultsECT significantly increased global efficiency, edge betweenness centrality, local efficiency, and mean degree of alpha band after ECT treatment, and an increase in these network metrics had significant correlations with decreased depressive symptoms in repeated measures correlation. In addition, ECT regulated the distribution of hubs in frontal and occipital lobes.ConclusionECT modulated the brain’s global and local information-processing patterns. In addition, an ECT-induced increase in network metrics was associated with clinical remission.SignificanceThese findings might present the evidence for us to understand how ECT regulated the topology organization in functional brain networks of clinically remitted depressive patients.

Brain Network Resilience: From MDD, Bipolar Disorder to Multiple Sclerosis

Stress is a major risk factor for the development of psychiatric and neurological disorders. An efficient recovery in the aftermath of stress is important for optimal resilience. Increased stress vulnerability across psychiatric and neurological disorders may therefore be related to altered trajectories during the recovery phase following stress, quantified by changes in functional brain networks.

Cannabis Use and Neuroadaptation: A Call for Δ 9 -Tetrahydrocannabinol Challenge Studies.

Currently, the assessment of the neurobehavioral consequences of repeated cannabis use is restricted to studies in which brain function of chronic cannabis users is compared to that of non-cannabis using controls. The assumption of such studies is that changes in brain function of chronic users are caused by repeated and prolonged exposure to acute cannabis intoxication. However, differences in brain function between chronic cannabis users and non-users might also arise from confounding factors such as polydrug use, alcohol use, withdrawal, economic status, or lifestyle conditions. We propose a methodology that highlights the relevance of acute Δ9-tetrahydrocannabinol (THC) dosing studies for a direct assessment of neuroadaptations in chronic cannabis users. The approach includes quantification of neurochemical, receptor, and functional brain network changes in response to an acute cannabis challenge, as well as stratification of cannabis using groups ranging from occasional to cannabis-dependent individuals. The methodology allows for an evaluation of THC induced neuroadaptive and neurocognitive changes across cannabis use history, that can inform neurobiological models on reward driven, compulsive cannabis use.

Brain Functional Network and Amino Acid Metabolism Association in Females with Subclinical Depression

This study aimed to investigate the association between complex brain functional networks and the metabolites in urine in subclinical depression. Electroencephalography (EEG) signals were recorded from 78 female college students, including 40 with subclinical depression (ScD) and 38 healthy controls (HC). The phase delay index was utilized to construct functional connectivity networks and quantify the topological properties of brain networks using graph theory. Meanwhile, the urine of all participants was collected for non-targeted LC-MS metabolic analysis to screen differential metabolites. The global efficiency was significantly increased in the α-2, β-1, and β-2 bands, while the characteristic path length of β-1 and β-2 and the clustering coefficient of β-2 were decreased in the ScD group. The severity of depression was negatively correlated with the level of cortisone (p = 0.016, r = −0.40). The metabolic pathways, including phenylalanine metabolism, phenylalanine tyrosine tryptophan biosynthesis, and nitrogen metabolism, were disturbed in the ScD group. The three metabolic pathways were negatively correlated (p = 0.014, r = −0.493) with the global efficiency of the brain network of the β-2 band, whereas they were positively correlated (p = 0.014, r = 0.493) with the characteristic path length of the β-2 band. They were mainly associated with low levels of L-phenylalanine, and the highest correlation sparsity was 0.11. The disturbance of phenylalanine metabolism and the phenylalanine, tryptophan, tyrosine biosynthesis pathways cause depressive symptoms and changes in functional brain networks. The decrease in the L-phenylalanine level may be related to the randomization trend of the β-1 frequency brain functional network.

EEG based functional brain networks analysis in dyslexic children during arithmetic task.

Developmental Dyslexia is a neuro-developmental disorder that often refers to a phonological processing deficit regardless of average IQ. The present study investigated the distinct functional changes in brain networks of dyslexic children during arithmetic task performance using an electroencephalogram. Fifteen dyslexic children and fifteen normally developing children (NDC) were recruited and performed an arithmetic task. Brain functional network measures such as node strength, clustering coefficient, characteristic pathlength and small-world were calculated using graph theory methods for both groups. Task performance showed significantly less performance accuracy in dyslexics against NDC. The neural findings showed increased connectivity in the delta band and reduced connectivity in theta, alpha, and beta band at temporoparietal, and prefrontal regions in dyslexic group while performing the task. The node strengths were found to be significantly high in delta band (T3, O1, F8 regions) and low in theta (T5, P3, Pz regions), beta (Pz) and gamma band (T4 and prefrontal regions) during the task in dyslexics compared to the NDC. The clustering coefficient was found to be significantly low in the dyslexic group (theta and alpha band) and characteristic pathlength was found to be significantly high in the dyslexic group (theta and alpha band) compared to the NDC group while performing task. In conclusion, the present study shows evidence for poor fact-retrieval mechanism and altered network topology in dyslexic brain networks during arithmetic task performance.

Imaging Clinical Subtypes and Associated Brain Networks in Alzheimer's Disease.

Alzheimer’s disease (AD) does not present uniform symptoms or a uniform rate of progression in all cases. The classification of subtypes can be based on clinical symptoms or patterns of pathological brain alterations. Imaging techniques may allow for the identification of AD subtypes and their differentiation from other neurodegenerative diseases already at an early stage. In this review, the strengths and weaknesses of current clinical imaging methods are described. These include positron emission tomography (PET) to image cerebral glucose metabolism and pathological amyloid or tau deposits. Magnetic resonance imaging (MRI) is more widely available than PET. It provides information on structural or functional changes in brain networks and their relation to AD subtypes. Amyloid PET provides a very early marker of AD but does not distinguish between AD subtypes. Regional patterns of pathology related to AD subtypes are observed with tau and glucose PET, and eventually as atrophy patterns on MRI. Structural and functional network changes occur early in AD but have not yet provided diagnostic specificity.

Electrophysiological resting state brain network and episodic memory in healthy aging adults

Recent studies have emphasized the changes in large-scale brain networks related to healthy aging, with the ultimate purpose to aid in differentiating normal neurocognitive aging from neurodegenerative disorders that also arise with age. Emerging evidence from functional Magnetic Resonance Imaging (fMRI) indicates that connectivity patterns within specific brain networks, especially the Default Mode Network (DMN), distinguish those with Alzheimer's disease from healthy individuals. In addition, disruptive alterations in the large-scale brain systems that support high-level cognition are shown to accompany cognitive decline at the behavioral level, which is commonly observed in the aging populations, even in the absence of disease. Although fMRI is useful for assessing functional changes in brain networks, its high costs and limited accessibility discourage studies that need large populations. In this study, we investigated the aging-effect on large-scale networks of the human brain using high-density electroencephalography and electrophysiological source imaging, which is a less costly and more accessible alternative to fMRI. In particular, our study examined a group of healthy subjects in the age range from middle- to older-aged adults, which is an under-studied range in the literature. Employing a high-resolution computation model, our results revealed age associations in the connectivity pattern of DMN in a consistent manner with previous fMRI findings. Particularly, in combination with a standard battery of cognitive tests, our data showed that in the posterior cingulate / precuneus area of DMN higher brain connectivity was associated with lower performance on an episodic memory task. The findings demonstrate the feasibility of using electrophysiological imaging to characterize large-scale brain networks and suggest that changes in network connectivity are associated with normal aging.

Oscillatory brain network changes after transcranial magnetic stimulation treatment in patients with major depressive disorder

• Oscillatory network changes during rTMS in MDD were evaluated by clinical EEG. • Functional connectivity changes were observed in MDD after rTMS. • Improved theta-band functional connectivity was associated with improved cognition. : Repetitive transcranial magnetic stimulation (rTMS) has attracted attention for treating treatment-resistant major depressive disorder (MDD) because of its effectiveness and low invasiveness. Recent neuroimaging studies suggest that the effects of rTMS in MDD may be based on improvements in abnormal brain networks. However, oscillatory network changes after rTMS remain unclear. This study aimed to detect oscillatory electroencephalography (EEG) markers that reflect the therapeutic effects of rTMS. : We treated 15 treatment-resistant MDD patients with high-frequency (10 Hz) rTMS over the left prefrontal cortex for 6 weeks. The power spectral density at each electrode and the phase lag index between all electrode pairs were compared before and after rTMS using clinical EEG data. The relationships between EEG indices, depressive symptoms, and cognitive function were also examined. : Depressive symptoms were significantly improved after rTMS. Additionally, functional connectivity between the left frontal–temporal and occipital regions in the theta band and gamma power in the anterior–central area were increased in both hemispheres after rTMS. Increased functional connectivity in the theta band was correlated with improved cognitive function, while increased gamma power was correlated with worsened depressive symptoms. : The sample size was relatively small. : Our results demonstrated oscillatory changes in theta-band functional connectivity after rTMS in MDD patients. These findings, obtained using clinical EEG methods, provide evidence of the underlying neurophysiological effects of rTMS on MDD within the context of functional brain network changes, and may constitute a biomarker for evaluating MDD treatments.

Associations of the Disrupted Functional Brain Network and Cognitive Function in End-Stage Renal Disease Patients on Maintenance Hemodialysis: A Graph Theory-Based Study of Resting-State Functional Magnetic Resonance Imaging.

Objective: Cognitive impairment (CI) is a common neurological complication in patients with end-stage renal disease undergoing maintenance hemodialysis (MHD). Brain network analysis based on graph theory is a promising tool for studying CI. Therefore, the purpose of this study was to analyze the changes of functional brain networks in patients on MHD with and without CI by using graph theory and further explore the underlying neuropathological mechanism of CI in these patients.Methods: A total of 39 patients on MHD (19 cases with CI and 20 without) and 25 healthy controls (HCs) matched for age, sex, and years of education were enrolled in the study. Resting-state functional magnetic resonance imaging (rs-fMRI) and T1-weighted high-resolution anatomical data were obtained, and functional brain networks for each subject were constructed. The brain network parameters at the global and regional levels were calculated, and a one-way analysis of covariance was used to compare the differences across the three groups. The associations between the changed graph-theory parameters and cognitive function scores in patients on MHD were evaluated using Spearman correlation analysis.Results: Compared with HCs, the global parameters [sigma, gamma, and local efficiency (Eloc)] in both patient groups decreased significantly (p < 0.05, Bonferroni corrected). The clustering coefficient (Cp) in patients with CI was significantly lower than that in the other two groups (p < 0.05, Bonferroni corrected). The regional parameters were significantly lower in the right superior frontal gyrus, dorsolateral (SFGdor) and gyrus rectus (REC) of patients with CI than those of patients without CI; however the nodal local efficiency in the left amygdala was significantly increased (all p < 0.05, Bonferroni corrected). The global Cp and regional parameters in the three brain regions (right SFGdor, REC, and left amygdala) were significantly correlated with the cognitive function scores (all FDR q < 0.05).Conclusion: This study confirmed that the topology of the functional brain network was disrupted in patients on MHD with and without CI and the disruption of brain network was more severe in patients with CI. The abnormal brain network parameters are closely related to cognitive function in patients on MHD.

Subthreshold amyloid pathology and changes in functional brain networks in patients with delayed cognitive decline after stroke

AbstractBackgroundIn clinical practice, some of the stroke survivors who have normal cognitive function until three to six months after index stroke, develop cognitive decline gradually after then. Although amyloid deposits and advanced small vessel disease are reported to be the possible predictors, pathogenesis for delayed cognitive decline has not been identified yet. Thus, we aimed to explore the brain network characteristics in those with delayed cognitive decline after stroke compared to those without.MethodUsing a prospective stroke registry, we identified patients with ischemic stroke who had undergone the cognitive evaluations including mini‐mental state examination (MMSE) twice or more after index stroke. We included patients whose baseline cognitive evaluation (between 1 and 12 months after stroke) was normal and who had one or more follow‐up evaluations after more than 6 months from the baseline evaluation. Among 222 eligible patients, decliners were defined as having a decline in 3 or more per year or an absolute decline of 5 or more between baseline and follow‐up MMSE, while non‐decliners as having a decline of 1 point per year or less and an absolute decline of less than 3 points. Brain network attributes were analyzed using resting‐state functional MRI, and superimposed amyloid positivity was assessed using 18F‐florbetaben PET.ResultEleven patients (5%, 11/222) met the criteria for decliners, and ten matched non‐decliners were enrolled. For about 5.7 years of follow‐up, the mean change of MMSE scores was ‐9.1 points in the decliners, while ‐0.2 point in the non‐decliners. Ten (10/11) decliners were amyloid‐negative on the amyloid PET, and the APOE e4 allele was identified in one decliner. However, the global SUVR value showed a negative correlation with the final MMSE scores as well as the amount of MMSE change. Normalized transitivity was lower in decliners compared to non‐decliners. Mean connectivity strengths were averaged based on the seven canonical functional networks, and the intra‐network connections within the frontoparietal network showed the considerable differences between groups. (Figure)ConclusionThis study suggests that alteration of brain networks may explain the pathogenesis of the delayed post‐stroke cognitive decline independent of amyloid PET positivity and APOE genotype.

Controlling extended criticality via modular connectivity

Criticality has been conjectured as an integral part of neuronal network dynamics. Operating at a critical threshold requires precise parameter tuning and a corresponding mechanism remains an open question. Recent studies have suggested that topological features observed in brain networks give rise to a Griffiths phase, leading to power-law scaling in brain activity dynamics and the operational benefits of criticality in an extended parameter region. Motivated by growing evidence of neural correlates of different states of consciousness, we investigate how topological changes affect the expression of a Griffiths phase. We analyze the activity decay in modular networks using a susceptible-infected-susceptible propagation model and find that we can control the extension of the Griffiths phase by altering intra- and intermodular connectivity. We find that by adjusting system parameters, we can counteract changes in critical behavior and maintain a stable critical region despite changes in network topology. Our results give insight into how structural network properties affect the emergence of a Griffiths phase and how its features are linked to established topological network metrics. We discuss how those findings could contribute to an understanding of the changes in functional brain networks.

Dynamic Network Analysis Demonstrates the Formation of Stable Functional Networks During Rule Learning.

Variations in the functional connectivity of large-scale cortical brain networks may explain individual differences in learning ability. We used a dynamic network analysis of fMRI data to identify changes in functional brain networks that are associated with context-dependent rule learning. During fMRI scanning, naïve subjects performed a cognitive task designed to test their ability to learn context-dependent rules. Notably, subjects were given minimal instructions about the task prior to scanning. We identified several key network characteristics associated with fast and accurate rule learning. First, consistent with the formation of stable functional networks, a dynamic community detection analysis revealed regionally specific reductions in flexible switching between different functional communities in successful learners. Second, successful rule learners showed decreased centrality of ventral attention regions and increased assortative mixing of cognitive control regions as the rules were learned. Finally, successful subjects showed greater decoupling of default and attention communities throughout the entire task, whereas ventral attention and cognitive control regions became more connected during learning. Overall, the results support a framework by which a stable ventral attention community and more flexible cognitive control community support sustained attention and the formation of rule representations in successful learners.

Learning dynamic graph embeddings for accurate detection of cognitive state changes in functional brain networks

Mounting evidence shows that brain functions and cognitive states are dynamically changing even in the resting state rather than remaining at a single constant state. Due to the relatively small changes in BOLD (blood-oxygen-level-dependent) signals across tasks, it is difficult to detect the change of cognitive status without requiring prior knowledge of the experimental design. To address this challenge, we present a dynamic graph learning approach to generate an ensemble of subject-specific dynamic graph embeddings, which allows us to use brain networks to disentangle cognitive events more accurately than using raw BOLD signals. The backbone of our method is essentially a representation learning process for projecting BOLD signals into a latent vertex-temporal domain with the greater biological underpinning of brain activities. Specifically, the learned representation domain is jointly formed by (1) a set of harmonic waves that govern the topology of whole-brain functional connectivities and (2) a set of Fourier bases that characterize the temporal dynamics of functional changes. In this regard, our dynamic graph embeddings provide a new methodology to investigate how these self-organized functional fluctuation patterns oscillate along with the evolving cognitive status. We have evaluated our proposed method on both simulated data and working memory task-based fMRI datasets, where our dynamic graph embeddings achieve higher accuracy in detecting multiple cognitive states than other state-of-the-art methods.

A Graph Gaussian Embedding Method for Predicting Alzheimer's Disease Progression With MEG Brain Networks.

Characterizing the subtle changes of functional brain networks associated with the pathological cascade of Alzheimer's disease (AD) is important for early diagnosis and prediction of disease progression prior to clinical symptoms. We developed a new deep learning method, termed multiple graph Gaussian embedding model (MG2G), which can learn highly informative network features by mapping high-dimensional resting-state brain networks into a low-dimensional latent space. These latent distribution-based embeddings enable a quantitative characterization of subtle and heterogeneous brain connectivity patterns at different regions, and can be used as input to traditional classifiers for various downstream graph analytic tasks, such as AD early stage prediction, and statistical evaluation of between-group significant alterations across brain regions. We used MG2G to detect the intrinsic latent dimensionality of MEG brain networks, predict the progression of patients with mild cognitive impairment (MCI) to AD, and identify brain regions with network alterations related to MCI.