Brain Dynamics Research Articles

Frontal theta oscillations during emotion regulation in people with borderline personality disorder.

Borderline personality disorder (BPD) is a severe psychiatric disorder conceptualised as a disorder of emotion regulation. Emotion regulation has been linked to a frontolimbic network comprising the dorsolateral prefrontal cortex and the amygdala, which apparently synchronises its activity via oscillatory coupling in the theta frequency range. To analyse whether there are distinct differences in theta oscillatory coupling in frontal brain regions between individuals with BPD and matched controls during emotion regulation by cognitive reappraisal. Electroencephalogram (EEG) recordings were performed in 25 women diagnosed with BPD and 25 matched controls during a cognitive reappraisal task in which participants were instructed to downregulate negative emotions evoked by aversive visual stimuli. Between- and within-group time-frequency analyses were conducted to analyse regulation-associated theta activity (3.5-8.5 Hz). Oscillatory theta activity differed between the participants with BPD and matched controls during cognitive reappraisal. Regulation-associated theta increases were lower in frontal regions in the BPD cohort compared with matched controls. Functional connectivity analysis for regulation-associated changes in the theta frequency band revealed a lower multivariate interaction measure (MIM) increase in frontal brain regions in persons with BPD compared with matched controls. Our findings support the notion of alterations in a frontal theta network in BPD, which may be underlying core symptoms of the disorder such as deficits in emotion regulation. The results add to the growing body of evidence for altered oscillatory brain dynamics in psychiatric populations, which might be investigated as individualised treatment targets using non-invasive stimulation methods.

A Multi-Graph Cross-Attention-Based Region-Aware Feature Fusion Network Using Multi-Template for Brain Disorder Diagnosis.

Functional connectivity (FC) networks based on resting-state functional magnetic imaging (rs-fMRI) are reliable and sensitive for brain disorder diagnosis. However, most existing methods are limited by using a single template, which may be insufficient to reveal complex brain connectivities. Furthermore, these methods usually neglect the complementary information between static and dynamic brain networks, and the functional divergence among different brain regions, leading to suboptimal diagnosis performance. To address these limitations, we propose a novel multi-graph cross-attention based region-aware feature fusion network (MGCA-RAFFNet) by using multi-template for brain disorder diagnosis. Specifically, we first employ multi-template to parcellate the brain space into different regions of interest (ROIs). Then, a multi-graph cross-attention network (MGCAN), including static and dynamic graph convolutions, is developed to explore the deep features contained in multi-template data, which can effectively analyze complex interaction patterns of brain networks for each template, and further adopt a dual-view cross-attention (DVCA) to acquire complementary information. Finally, to efficiently fuse multiple static-dynamic features, we design a region-aware feature fusion network (RAFFNet), which is beneficial to improve the feature discrimination by considering the underlying relations among static-dynamic features in different brain regions. Our proposed method is evaluated on both public ADNI-2 and ABIDE-I datasets for diagnosing mild cognitive impairment (MCI) and autism spectrum disorder (ASD). Extensive experiments demonstrate that the proposed method outperforms the state-of-the-art methods. Our source code is available at https://github.com/mylbuaa/MGCA-RAFFNet.

Information-Theoretical Analysis of the Cycle of Creation of Knowledge and Meaning in Brains under Multiple Cognitive Modalities.

It is of great interest to develop advanced sensory technologies allowing non-invasive monitoring of neural correlates of cognitive processing in people performing everyday tasks. A lot of progress has been reported in recent years in this research area using scalp EEG arrays, but the high level of noise in the electrode signals poses a lot of challenges. This study presents results of detailed statistical analysis of experimental data on the cycle of creation of knowledge and meaning in human brains under multiple cognitive modalities. We measure brain dynamics using a HydroCel Geodesic Sensor Net, 128-electrode dense-array electroencephalography (EEG). We compute a pragmatic information (PI) index derived from analytic amplitude and phase, by Hilbert transforming the EEG signals of 20 participants in six modalities, which combine various audiovisual stimuli, leading to different mental states, including relaxed and cognitively engaged conditions. We derive several relevant measures to classify different brain states based on the PI indices. We demonstrate significant differences between engaged brain states that require sensory information processing to create meaning and knowledge for intentional action, and relaxed-meditative brain states with less demand on psychophysiological resources. We also point out that different kinds of meanings may lead to different brain dynamics and behavioral responses.

A general description of criticality in neural network models

Recent experimental observations have supported the hypothesis that the cerebral cortex operates in a dynamical regime near criticality, where the neuronal network exhibits a mixture of ordered and disordered patterns. However, A comprehensive study of how criticality emerges and how to reproduce it is still lacking. In this study, we investigate coupled networks with conductance-based neurons and illustrate the co-existence of different spiking patterns, including asynchronous irregular (AI) firing and synchronous regular (SR) state, along with a scale-invariant neuronal avalanche phenomenon (criticality). We show that fast-acting synaptic coupling can evoke neuronal avalanches in the mean-dominated regime but has little effect in the fluctuation-dominated regime. In a narrow region of parameter space, the network exhibits avalanche dynamics with power-law avalanche size and duration distributions. We conclude that three stages which may be responsible for reproducing the synchronized bursting: mean-dominated subthreshold dynamics, fast-initiating a spike event, and time-delayed inhibitory cancellation. Remarkably, we illustrate the mechanisms underlying critical avalanches in the presence of noise, which can be explained as a stochastic crossing state around the Hopf bifurcation under the mean-dominated regime. Moreover, we apply the ensemble Kalman filter to determine and track effective connections for the neuronal network. The method is validated on noisy synthetic BOLD signals and could exactly reproduce the corresponding critical network activity. Our results provide a special perspective to understand and model the criticality, which can be useful for large-scale modeling and computation of brain dynamics.

Sympathetic and parasympathetic central autonomic networks

Abstract The central-autonomic network (CAN) comprises brain regions that are functionally linked to the activity of peripheral autonomic nerves. While parasympathetic CAN (i.e., the CAN projecting onto parasympathetic branches) has recently been investigated and is known to be involved in neurological and neuropsychiatric disorders, sympathetic CAN (i.e., the CAN projecting onto sympathetic nerves) has not been fully characterized. Using functional magnetic resonance imaging (fMRI) data from the Human Connectome Project in conjunction with heartbeat dynamics and its orthonormal autoregressive descriptors as a proxy for sympathetic activity estimation, namely, the sympathetic activity index (SAI), we uncover brain regions belonging to the sympathetic CAN at rest. We uncover a widespread CAN comprising both cortical (in all lobes) and subcortical areas, including the cerebellum and brainstem, which is functionally linked to sympathetic activity and overlaps with brain regions driving parasympathetic activity. These findings may constitute fundamental knowledge linking brain and bodily dynamics, including the link between neurological and psychiatric disorders and autonomic dysfunctions.

Neuronal dynamics direct cerebrospinal fluid perfusion and brain clearance.

The accumulation of metabolic waste is a leading cause of numerous neurological disorders, yet we still have only limited knowledge of how the brain performs self-cleansing. Here we demonstrate that neural networks synchronize individual action potentials to create large-amplitude, rhythmic and self-perpetuating ionic waves in the interstitial fluid of the brain. These waves are a plausible mechanism to explain the correlated potentiation of the glymphatic flow1,2 through the brain parenchyma. Chemogenetic flattening of these high-energy ionic waves largely impeded cerebrospinal fluid infiltration into and clearance of molecules from the brain parenchyma. Notably, synthesized waves generated through transcranial optogenetic stimulation substantially potentiated cerebrospinal fluid-to-interstitial fluid perfusion. Our study demonstrates that neurons serve as master organizers for brain clearance. This fundamental principle introduces a new theoretical framework for the functioning of macroscopic brain waves.

Changes in Community Structure of Brain Dynamic Functional Connectivity States in Mild Cognitive Impairment

Recent researches have noted many changes of short-term dynamic modalities in mild cognitive impairment (MCI) patients' brain functional networks. In this study, the dynamic functional brain networks of 82 MCI patients and 85 individuals in the normal control (NC) group were constructed using the sliding window method and Pearson correlation. The window size was determined using single-scale time-dependent (SSTD) method. Subsequently, k-means was applied to cluster all window samples, identifying three dynamic functional connectivity (DFC) states. Collective sparse symmetric non-negative matrix factorization (cssNMF) was then used to perform community detection on these states and quantify differences in brain regions. Finally, metrics such as within-community connectivity strength, community strength, and node diversity were calculated for further analysis. The results indicated high similarity between the two groups in state 2, with no significant differences in optimal community quantity and functional segregation (p < 0.05). However, for state 1 and state 3, the optimal community quantity was smaller in MCI patients compared to the NC group. In state 1, MCI patients had lower within-community connectivity strength and overall strength than the NC group, whereas state 3 showed results opposite to state 1. Brain regions with statistical difference included MFG.L, ORBinf.R, STG.R, IFGtriang.L, CUN.L, CUN.R, LING.R, SOG.L, and PCUN.R. This study on DFC states explores changes in the brain functional networks of patients with MCI from the perspective of alterations in the community structures of DFC states. The findings could provide new insights into the pathological changes in the brains of MCI patients.

Freezing Effect and Bystander Effect: Overlaps and Differences

The present article provides a detailed comparison of two psychological phenomena, the freezing effect and the bystander effect, across their neurobiological, cognitive, emotional, and behavioral dimensions. This study focuses on identifying and analyzing the similarities and differences between these two responses to stressful and traumatic events. While the freezing effect is characterized by an involuntary neurobiological response to immediate threats, resulting in temporary immobilization or paralysis, the bystander effect describes a cognitive and social phenomenon where individuals refrain from offering help in emergencies when others are present. The study explores affective aspects, including emotional responses and trauma-related impacts associated with both phenomena. Through a comparative analysis, this research unveils important understandings regarding the distinctions among these responses, including their triggers, underlying mechanisms, and observable behaviors. It also highlights overlapping aspects, particularly in how both phenomena can lead to inaction in critical moments. This comparative study contributes to a deeper understanding of the complex interaction between the brain, individual cognition, and social dynamics in the face of danger and stress. The findings of this research have significant implications for understanding human behavior in emergencies, offering valuable perspectives that can be applied in the domains of psychology, training for emergency response, and trauma therapy.

Distinct neocortical mechanisms underlie human SI responses to median nerve and laser-evoked peripheral activation

Abstract Magneto- and/or electro-encephalography (M/EEG) are non-invasive clinically relevant tools that have long been used to measure electromagnetic fields in the somatosensory cortex evoked by innocuous and noxious somatosensory stimuli. Two commonly applied stimulation paradigms that produce distinct responses in the primary somatosensory cortex (SI) linked to innocuous and noxious sensations are electrical median nerve (MN) stimulation and cutaneous laser-evoked (LE) stimulation to the dorsum of the hand, respectively. Despite their prevalence, the physiological mechanisms that produce stereotypic macroscale MN and LE responses have yet to be fully articulated, limiting their utility in understanding brain dynamics associated with non-painful and/or painful somatosensation. Through a literature review, we detailed features of MN and LE responses source-localized to SI that are robust and reproducible across studies. We showed that the first peak in the MN response at ~20 ms post-stimulus (i.e., MN N1) corresponds to upward-directed deep-to-superficial electrical current flow through the cortical laminae, which is followed by downward-directed current at ~30 ms (i.e., MN P1). In contrast, the initial LE response occurs later at ~170 ms (i.e., LE N1) and is directed downward and opposite the direction of the MN N1. We then examined the neocortical circuit mechanisms contributing to the robust features of each response using the Human Neocortical Neurosolver (HNN) neural modeling software tool (Neymotin et al., 2020). Using HNN as a hypothesis development and testing tool, model results predicted the MN response can be simulated with a sequence of layer-specific thalamocortical and cortico-cortical synaptic drive similar to that previously reported for tactile evoked responses (S. R. Jones et al., 2007; Neymotin et al., 2020), with the novel discovery that an early excitatory input to supragranular layers at ~30 ms is an essential mechanism contributing to the downward current flow of the MN P1. Model results further predicted that the initial ~170 ms downward current flow of the LE N1 was generated by a burst of repetitive gamma-frequency (~40 Hz) excitatory synaptic drive to supragranular layers, consistent with prior reports of LE gamma-frequency activity. These results make novel and detailed multiscale predictions about the dynamic laminar circuit mechanisms underlying temporal and spectral features of MN and LE responses in SI and can guide further investigations in follow-up studies. Ultimately, these findings may help with the development of targeted therapeutics for pathological somatosensation, such as somatic sensitivity and acute neuropathic pain.

Deep learning models reveal replicable, generalizable, and behaviorally relevant sex differences in human functional brain organization

Sex plays a crucial role in human brain development, aging, and the manifestation of psychiatric and neurological disorders. However, our understanding of sex differences in human functional brain organization and their behavioral consequences has been hindered by inconsistent findings and a lack of replication. Here, we address these challenges using a spatiotemporal deep neural network (stDNN) model to uncover latent functional brain dynamics that distinguish male and female brains. Our stDNN model accurately differentiated male and female brains, demonstrating consistently high cross-validation accuracy (>90%), replicability, and generalizability across multisession data from the same individuals and three independent cohorts (N ~ 1,500 young adults aged 20 to 35). Explainable AI (XAI) analysis revealed that brain features associated with the default mode network, striatum, and limbic network consistently exhibited significant sex differences (effect sizes > 1.5) across sessions and independent cohorts. Furthermore, XAI-derived brain features accurately predicted sex-specific cognitive profiles, a finding that was also independently replicated. Our results demonstrate that sex differences in functional brain dynamics are not only highly replicable and generalizable but also behaviorally relevant, challenging the notion of a continuum in male-female brain organization. Our findings underscore the crucial role of sex as a biological determinant in human brain organization, have significant implications for developing personalized sex-specific biomarkers in psychiatric and neurological disorders, and provide innovative AI-based computational tools for future research.

A Small-Molecule Ratiometric Photoacoustic Probe for the High-Spatiotemporal-Resolution Imaging of Copper(II) Dynamics in the Mouse Brain.

Copper dysmetabolism is associated with various neurodegenerative disorders, making high-spatiotemporal-resolution imaging of Cu2+ in the brain essential for understanding the underlying pathophysiological processes. Nevertheless, the current probes encounter obstacles in crossing the blood-brain barrier (BBB) and providing high-spatial-resolution in deep tissues. Herein, we present a photoacoustic probe capable of imaging Cu2+ dynamics in the mouse brain with high-spatiotemporal-resolution. The probe demonstrates selective ratiometric and reversible responses to Cu2+ , while also efficiently crossing the BBB. Using the probe as the imaging agent, we successfully visualized Cu2+ in the brain of Parkinson's disease (PD) model mouse with a remarkable micron-level resolution. The imaging results revealed a significant increase in Cu2+ levels in the cerebral cortex as PD progresses, highlighting the close association between Cu2+ alternations in the region and the disease. We also demonstrated that the probe can be used to monitor changes in Cu2+ distribution in the PD model mouse brain during L-dopa intervention. Mechanism studies suggest that the copper dyshomeostasis in the PD mouse brain was dominated by the expression levels of divalent metal transporter 1. The application of our probe in imaging Cu2+ dynamics in the mouse brain offers valuable insights into the copper-related molecular mechanisms underlying neurodegenerative diseases.

A Small‐Molecule Ratiometric Photoacoustic Probe for the High‐Spatiotemporal‐Resolution Imaging of Copper(II) Dynamics in the Mouse Brain

AbstractCopper dysmetabolism is associated with various neurodegenerative disorders, making high‐spatiotemporal‐resolution imaging of Cu2+ in the brain essential for understanding the underlying pathophysiological processes. Nevertheless, the current probes encounter obstacles in crossing the blood–brain barrier (BBB) and providing high‐spatial‐resolution in deep tissues. Herein, we present a photoacoustic probe capable of imaging Cu2+ dynamics in the mouse brain with high‐spatiotemporal‐resolution. The probe demonstrates selective ratiometric and reversible responses to Cu2+, while also efficiently crossing the BBB. Using the probe as the imaging agent, we successfully visualized Cu2+ in the brain of Parkinson's disease (PD) model mouse with a remarkable micron‐level resolution. The imaging results revealed a significant increase in Cu2+ levels in the cerebral cortex as PD progresses, highlighting the close association between Cu2+ alternations in the region and the disease. We also demonstrated that the probe can be used to monitor changes in Cu2+ distribution in the PD model mouse brain during L‐dopa intervention. Mechanism studies suggest that the copper dyshomeostasis in the PD mouse brain was dominated by the expression levels of divalent metal transporter 1. The application of our probe in imaging Cu2+ dynamics in the mouse brain offers valuable insights into the copper‐related molecular mechanisms underlying neurodegenerative diseases.

Brain dynamics predictive of response to psilocybin for treatment-resistant depression.

Psilocybin therapy for depression has started to show promise, yet the underlying causal mechanisms are not currently known. Here, we leveraged the differential outcome in responders and non-responders to psilocybin (10 and 25 mg, 7 days apart) therapy for depression-to gain new insights into regions and networks implicated in the restoration of healthy brain dynamics. We used large-scale brain modelling to fit the spatiotemporal brain dynamics at rest in both responders and non-responders before treatment. Dynamic sensitivity analysis of systematic perturbation of these models enabled us to identify specific brain regions implicated in a transition from a depressive brain state to a healthy one. Binarizing the sample into treatment responders (>50% reduction in depressive symptoms) versus non-responders enabled us to identify a subset of regions implicated in this change. Interestingly, these regions correlate with in vivo density maps of serotonin receptors 5-hydroxytryptamine 2a and 5-hydroxytryptamine 1a, which psilocin, the active metabolite of psilocybin, has an appreciable affinity for, and where it acts as a full-to-partial agonist. Serotonergic transmission has long been associated with depression, and our findings provide causal mechanistic evidence for the role of brain regions in the recovery from depression via psilocybin.

Enzymatic Galvanic Redox Potentiometry for In Vivo Biosensing.

Redox potentiometry has emerged as a new platform for in vivo sensing, with improved neuronal compatibility and strong tolerance against sensitivity variation caused by protein fouling. Although enzymes show great possibilities in the fabrication of selective redox potentiometry, the fabrication of an enzyme electrode to output open-circuit voltage (EOC) with fast response remains challenging. Herein, we report a concept of novel enzymatic galvanic redox potentiometry (GRP) with improved time response coupling the merits of the high selectivity of enzyme electrodes with the excellent biocompatibility and reliability of GRP sensors. With a glucose biosensor as an illustration, we use flavin adenine dinucleotide-dependent glucose dehydrogenase as the recognition element and carbon black as the potential relay station to improve the response time. We find that the enzymatic GRP biosensor rapidly responds to glucose with a good linear relationship between EOC and the logarithm of glucose concentration within a range from 100 μM to 2.65 mM. The GRP biosensor shows high selectivity over O2 and coexisting neurochemicals, good reversibility, and sensitivity and can in vivo monitor glucose dynamics in rat brain. We believe that this study will pave a new platform for the in vivo potentiometric biosensing of chemical events with high reliability.

Healthy aging alters the oscillatory dynamics and fronto-parietal connectivity serving fluid intelligence.

Fluid intelligence (Gf) involves logical reasoning and novel problem-solving abilities. Often, abstract reasoning tasks like Raven's progressive matrices are used to assess Gf. Prior work has shown an age-related decline in fluid intelligence capabilities, and although many studies have sought to identify the underlying mechanisms, our understanding of the critical brain regions and dynamics remains largely incomplete. In this study, we utilized magnetoencephalography (MEG) to investigate 78 individuals, ages 20-65 years, as they completed an abstract reasoning task. MEG data was co-registered with structural MRI data, transformed into the time-frequency domain, and the resulting neural oscillations were imaged using a beamformer. We found worsening behavioral performance with age, including prolonged reaction times and reduced accuracy. MEG analyses indicated robust oscillations in the theta, alpha/beta, and gamma range during the task. Whole brain correlation analyses with age revealed relationships in the theta and alpha/beta frequency bands, such that theta oscillations became stronger with increasing age in a right prefrontal region and alpha/beta oscillations became stronger with increasing age in parietal and right motor cortices. Follow-up connectivity analyses revealed increasing parieto-frontal connectivity with increasing age in the alpha/beta frequency range. Importantly, our findings are consistent with the parieto-frontal integration theory of intelligence (P-FIT). These results further suggest that as people age, there may be alterations in neural responses that are spectrally specific, such that older people exhibit stronger alpha/beta oscillations across the parieto-frontal network during abstract reasoning tasks.

A dynamical computational model of theta generation in hippocampal circuits to study theta-gamma oscillations during neurostimulation.

Neurostimulation of the hippocampal formation has shown promising results for modulating memory but the underlying mechanisms remain unclear. In particular, the effects on hippocampal theta-nested gamma oscillations and theta phase reset, which are both crucial for memory processes, are unknown. Moreover, these effects cannot be investigated using current computational models, which consider theta oscillations with a fixed amplitude and phase velocity. Here, we developed a novel computational model that includes the medial septum, represented as a set of abstract Kuramoto oscillators producing a dynamical theta rhythm with phase reset, and the hippocampal formation, composed of biophysically realistic neurons and able to generate theta-nested gamma oscillations under theta drive. We showed that, for theta inputs just below the threshold to induce self-sustained theta-nested gamma oscillations, a single stimulation pulse could switch the network behavior from non-oscillatory to a state producing sustained oscillations. Next, we demonstrated that, for a weaker theta input, pulse train stimulation at the theta frequency could transiently restore seemingly physiological oscillations. Importantly, the presence of phase reset influenced whether these two effects depended on the phase at which stimulation onset was delivered, which has practical implications for designing neurostimulation protocols that are triggered by the phase of ongoing theta oscillations. This novel model opens new avenues for studying the effects of neurostimulation on the hippocampal formation. Furthermore, our hybrid approach that combines different levels of abstraction could be extended in future work to other neural circuits that produce dynamical brain rhythms.

A classification-based generative approach to selective targeting of global slow oscillations during sleep.

Given sleep's crucial role in health and cognition, numerous sleep-based brain interventions are being developed, aiming to enhance cognitive function, particularly memory consolidation, by improving sleep. Research has shown that Transcranial Alternating Current Stimulation (tACS) during sleep can enhance memory performance, especially when used in a closed-loop (cl-tACS) mode that coordinates with sleep slow oscillations (SOs, 0.5-1.5Hz). However, sleep tACS research is characterized by mixed results across individuals, which are often attributed to individual variability. This study targets a specific type of SOs, widespread on the electrode manifold in a short delay ("global SOs"), due to their close relationship with long-term memory consolidation. We propose a model-based approach to optimize cl-tACS paradigms, targeting global SOs not only by considering their temporal properties but also their spatial profile. We introduce selective targeting of global SOs using a classification-based approach. We first estimate the current elicited by various stimulation paradigms, and optimize parameters to match currents found in natural sleep during a global SO. Then, we employ an ensemble classifier trained on sleep data to identify effective paradigms. Finally, the best stimulation protocol is determined based on classification performance. Our study introduces a model-driven cl-tACS approach that specifically targets global SOs, with the potential to extend to other brain dynamics. This method establishes a connection between brain dynamics and stimulation optimization. Our research presents a novel approach to optimize cl-tACS during sleep, with a focus on targeting global SOs. This approach holds promise for improving cl-tACS not only for global SOs but also for other physiological events, benefiting both research and clinical applications in sleep and cognition.

Combined effects of neuroticism and negative emotional context on spontaneous EEG dynamics.

Neuroticism is a personality trait with great clinical relevance, defined as a tendency to experience negative affect, sustained self-generated negative thoughts and impaired emotion regulation. Here, we investigated spontaneous brain dynamics in the aftermath of negative emotional events and their links with neuroticism in order to shed light on the prolonged activity of large-scale brain networks associated with the control of affect. We recorded electroencephalography (EEG) from 36 participants who were asked to rest after watching neutral or fearful video clips. Four topographic maps (i.e. microstates classes A, B, C and D) explained the majority of the variance in spontaneous EEG. Participants showed greater presence of microstate D and lesser presence of microstate C following exposure to fearful stimuli, pointing to changes in attention- and introspection-related networks previously associated with these microstates. These emotional effects were more pronounced for participants with low neuroticism. Moreover, neuroticism scores were positively correlated with microstate C and negatively correlated with microstate D, regardless of previous emotional stimulation. Our results reveal distinctive effects of emotional context on resting-state EEG, consistent with a prolonged impact of negative affect on the brain, and suggest a possible link with neuroticism.

Neonatal brain dynamic functional connectivity in term and preterm infants and its association with early childhood neurodevelopment

Brain dynamic functional connectivity characterises transient connections between brain regions. Features of brain dynamics have been linked to emotion and cognition in adult individuals, and atypical patterns have been associated with neurodevelopmental conditions such as autism. Although reliable functional brain networks have been consistently identified in neonates, little is known about the early development of dynamic functional connectivity. In this study we characterise dynamic functional connectivity with functional magnetic resonance imaging (fMRI) in the first few weeks of postnatal life in term-born (n = 324) and preterm-born (n = 66) individuals. We show that a dynamic landscape of brain connectivity is already established by the time of birth in the human brain, characterised by six transient states of neonatal functional connectivity with changing dynamics through the neonatal period. The pattern of dynamic connectivity is atypical in preterm-born infants, and associated with atypical social, sensory, and repetitive behaviours measured by the Quantitative Checklist for Autism in Toddlers (Q-CHAT) scores at 18 months of age.

Joint learning of multi-level dynamic brain networks for autism spectrum disorder diagnosis

Graph convolutional networks (GCNs), with their powerful ability to model non-Euclidean graph data, have shown advantages in learning representations of brain networks. However, considering the complexity, multilayeredness, and spatio-temporal dynamics of brain activities, we have identified two limitations in current GCN-based research on brain networks: 1) Most studies have focused on unidirectional information transmission across brain network levels, neglecting joint learning or bidirectional information exchange among networks. 2) Most of the existing models determine node neighborhoods by thresholding or simply binarizing the brain network, which leads to the loss of edge weight information and weakens the model's sensitivity to important information in the brain network. To address the above issues, we propose a multi-level dynamic brain network joint learning architecture based on GCN for autism spectrum disorder (ASD) diagnosis. Specifically, firstly, constructing different-level dynamic brain networks. Then, utilizing joint learning based on GCN for interactive information exchange among these multi-level brain networks. Finally, designing an edge self-attention mechanism to assign different edge weights to inter-node connections, which allows us to pick out the crucial features for ASD diagnosis. Our proposed method achieves an accuracy of 81.5 %. The results demonstrate that our method enables bidirectional transfer of high-order and low-order information, facilitating information complementarity between different levels of brain networks. Additionally, the use of edge weights enhances the representation capability of ASD-related features.