Group-PCA for very large fMRI datasets

Physiological Confounds in BOLD-fMRI and Their Correction

Functional magnetic resonance imaging (fMRI) has opened new frontiers in neuroscience by instrumentally driving our understanding of brain function and development. Despite its substantial successes, fMRI studies persistently encounter obstacles stemming from inherent, unavoidable physiological confounds. The adverse effects of these confounds are especially noticeable with higher magnetic fields, which have been gaining momentum in fMRI experiments. This review focuses on the four major physiological confounds impacting fMRI studies: low-frequency fluctuations in both breathing depth and rate, low-frequency fluctuations in the heart rate, thoracic movements, and cardiac pulsatility. Over the past three decades, numerous correction techniques have emerged to address these challenges. Correction methods have effectively enhanced the detection of task-activated voxels and minimized the occurrence of false positives and false negatives in functional connectivity studies. While confound correction methods have merit, they also have certain limitations. For instance, model-based approaches require externally recorded physiological data that is often unavailable in fMRI studies. Methods reliant on independent component analysis, on the other hand, need prior knowledge about the number of components. Machine learning techniques, although showing potential, are still in the early stages of development and require additional validation. This article reviews the mechanics of physiological confound correction methods, scrutinizes their performance and limitations, and discusses their impact on fMRI studies.

Anatomical heterogeneity in low-grade and high-grade gliomas: A multiscale perspective

BACKGROUND

Low-grade gliomas (LGGs) and high-grade gliomas (HGGs) often exhibit distinct spatial distributions, a phenomenon that remains incompletely understood. Based on previous research, we hypothesized that functional networks, neurotransmitters, and isocitrate dehydrogenase-1 (IDH-1) status characterize the spatial patterns of LGG and HGG.

METHODS

We analyzed 399 patients diagnosed with primary gliomas. First, we generated glioma frequency maps based on tumor grade, neurotransmitters, and IDH-1 status and constructed a brain functional connectivity network to explore heterogeneity in glioma location. Second, all tumor masks were mirror-symmetrized onto the brain's left hemisphere to facilitate feature extraction. We performed independent component analysis on merged four-dimensional files using Multivariate Exploratory Linear Optimized Decomposition into Independent Component (MELODIC), identifying four IDH-1 wild-type lesion covariance networks (IDHwt-LCNs) and three IDH-1 mutant lesion covariance networks (IDHmut-LCNs) with distinct spatial distributions, and analyzing correlation between the neurotransmitter levels and the IDH-wt/mut specific LCNs. Finally, we compared 42 white matter fibers extracted using XTRACT with 39 functional brain connectivity networks from the multi-subject dictionary learning (MSDL) atlas, revealing significant associations among the frontal aslant tract (FAT) and the intraparietal sulcus (IPS).

RESULTS

Our findings revealed high anatomical heterogeneity between LGG and HGG. Moreover, the high node strength played a critical role in the distinct spatial distribution of glioma. Significant correlations were observed between glioma frequency maps and dopaminergic, cholinergic, μ-opioid, and serotonergic neurotransmission. Furthermore, IDHwt/mut-LCNs analysis demonstrated that IDH-1 status influences glioma distribution, involving key brain structures. Lastly, we also found significant correlations between IDHwt/mut-LCNs and the neurotransmission of dopaminergic, cholinergic, μ-opioid, and serotonergic systems.

CONCLUSION

Our study highlighted the mechanisms by which functional networks, neurotransmitter systems, and IDH-1 status collectively contribute to the anatomical heterogeneity observed in LGG and HGG.

Manifold Learning Uncovers Nonlinear Interactions Between the Adolescent Brain and Environment That Predict Emotional and Behavioral Problems

BACKGROUND

To progress adolescent mental health research beyond our present achievements-a complex account of brain and environmental risk factors without understanding neurobiological embedding in the environment-we need methods to uncover relationships between the developing brain and real-world environmental experiences.

METHODS

We investigated associations between brain function, environments, and emotional and behavioral problems using participants from the Adolescent Brain Cognitive Development (ABCD) Study (n = 2401 female). We applied manifold learning, a promising technique for uncovering latent structure from high-dimensional biomedical data such as functional magnetic resonance imaging. Specifically, we developed exogenous PHATE (potential of heat-diffusion for affinity-based trajectory embedding) (E-PHATE) to model brain-environment interactions. We used E-PHATE embeddings of participants' brain activation during emotional and cognitive processing tasks to predict individual differences in cognition and emotional and behavioral problems both cross-sectionally and longitudinally.

RESULTS

E-PHATE embeddings of participants' brain activation and environments at baseline showed moderate-to-large associations with total, externalizing, and internalizing problems at baseline, across several subcortical regions and large-scale cortical networks, compared with the zero-to-small effects achieved by voxelwise data or common low-dimensional embedding methods. E-PHATE embeddings of the brain and environment at baseline were also related to emotional and behavioral problems 2 years later. These longitudinal predictions showed a consistent moderate effect in the frontoparietal and attention networks.

CONCLUSIONS

The embedding of the adolescent brain in the environment yields enriched insight into emotional and behavioral problems. Using E-PHATE, we demonstrated how the harmonization of cutting-edge computational methods with longstanding developmental theories advances the detection and prediction of adolescent emotional and behavioral problems.

A network correspondence toolbox for quantitative evaluation of novel neuroimaging results

The brain can be decomposed into large-scale functional networks, but the specific spatial topographies of these networks and the names used to describe them vary across studies. Such discordance has hampered interpretation and convergence of research findings across the field. We have developed the Network Correspondence Toolbox (NCT) to permit researchers to examine and report spatial correspondence between their novel neuroimaging results and multiple widely used functional brain atlases. We provide several exemplar demonstrations to illustrate how researchers can use the NCT to report their own findings. The NCT provides a convenient means for computing Dice coefficients with spin test permutations to determine the magnitude and statistical significance of correspondence among user-defined maps and existing atlas labels. The adoption of the NCT will make it easier for network neuroscience researchers to report their findings in a standardized manner, thus aiding reproducibility and facilitating comparisons between studies to produce interdisciplinary insights.

Distinct impact modes of polygenic disposition to dyslexia in the adult brain

Dyslexia is a common and partially heritable condition that affects reading ability. In a study of up to 35,231 adults, we explored the structural brain correlates of genetic disposition to dyslexia. Individual dyslexia-disposing genetic variants showed distinct patterns of association with brain structure. Independent component analysis revealed various brain networks that each had their own genomic profiles related to dyslexia susceptibility. Circuits involved in motor coordination, vision, and language were implicated. Polygenic scores for eight traits genetically correlated with dyslexia, including cognitive, behavioral, and reading-related psychometric measures, showed partial similarities to dyslexia in terms of brain-wide associations. Notably, microstructure of the internal capsule was consistently implicated across all of these genetic dispositions, while lower volume of the motor cortex was more specifically associated with dyslexia genetic disposition alone. These findings reveal genetic and neurobiological features that may contribute to dyslexia and its associations with other traits at the population level.

Brain functional gradients are related to cortical folding gradient

Cortical folding is closely linked to brain functions, with gyri acting more like local functional "hubs" to integrate information than sulci do. However, understanding how anatomical constraints relate to complex functions remains fragmented. One possible reason is that the relationship is estimated on brain mosaics divided by brain functions and cortical folding patterns. The boundaries of these hypothetical hard-segmented mosaics could be subject to the selection of functional/morphological features and as well as the thresholds. In contrast, functional gradient and folding gradient could provide a more feasible and unitless platform to mitigate the uncertainty introduced by boundary definition. Based on the MRI datasets, we used cortical surface curvature as the folding gradient and related it to the functional connectivity transition gradient. We found that, at the local scale, the functional gradient exhibits different function transition patterns between convex/concave cortices, with positive/negative curvatures, respectively. At the global scale, a cortex with more positive curvature could provide more function transition efficiency and play a more dominant role in more abstractive functional networks. These results reveal a novel relation between cortical morphology and brain functions, providing new clues to how anatomical constraint is related to the rise of an efficient brain function architecture.

Resting state of human brain measured by fMRI experiment is governed more dominantly by essential mode as a global signal rather than default mode network

Resting-state of the human brain has been described by a combination of various basis modes including the default mode network (DMN) identified by fMRI BOLD signals in human brains. Whether DMN is the most dominant representation of the resting-state has been under question. Here, we investigated the unexplored yet fundamental nature of the resting-state. In the absence of global signal regression for the analysis of brain-wide spatial activity pattern, the fMRI BOLD spatiotemporal signals during the rest were completely decomposed into time-invariant spatial-expression basis modes (SEBMs) and their time-evolution basis modes (TEBMs). Contrary to our conventional concept above, similarity clustering analysis of the SEBMs from 166 human brains revealed that the most dominant SEBM cluster is an asymmetric mode where the distribution of the sign of the components is skewed in one direction, for which we call essential mode (EM), whereas the second dominant SEBM cluster resembles the spatial pattern of DMN. Having removed the strong 1/f noise in the power spectrum of TEBMs, the genuine oscillatory behavior embedded in TEBMs of EM and DMN-like mode was uncovered around the low-frequency range below 0.2 Hz.

High-level cognition is supported by information-rich but compressible brain activity patterns

To efficiently yet reliably represent and process information, our brains need to produce information-rich signals that differentiate between moments or cognitive states, while also being robust to noise or corruption. For many, though not all, natural systems, these two properties are often inversely related: More information-rich signals are less robust, and vice versa. Here, we examined how these properties change with ongoing cognitive demands. To this end, we applied dimensionality reduction algorithms and pattern classifiers to functional neuroimaging data collected as participants listened to a story, temporally scrambled versions of the story, or underwent a resting state scanning session. We considered two primary aspects of the neural data recorded in these different experimental conditions. First, we treated the maximum achievable decoding accuracy across participants as an indicator of the "informativeness" of the recorded patterns. Second, we treated the number of features (components) required to achieve a threshold decoding accuracy as a proxy for the "compressibility" of the neural patterns (where fewer components indicate greater compression). Overall, we found that the peak decoding accuracy (achievable without restricting the numbers of features) was highest in the intact (unscrambled) story listening condition. However, the number of features required to achieve comparable classification accuracy was also lowest in the intact story listening condition. Taken together, our work suggests that our brain networks flexibly reconfigure according to ongoing task demands and that the activity patterns associated with higher-order cognition and high engagement are both more informative and more compressible than the activity patterns associated with lower-order tasks and lower engagement.

Multiscale Modes of Functional Brain Connectivity

Information processing in the brain spans from localised sensorimotor processes to higher-level cognition that integrates across multiple regions. Interactions between and within these subsystems enable multiscale information processing. Despite this multiscale characteristic, functional brain connectivity is often either estimated based on 10-30 distributed modes or parcellations with 100-1000 localised parcels, both missing across-scale functional interactions. We present Multiscale Probabilistic Functional Modes (mPFMs), a new mapping which comprises modes over various scales of granularity, thus enabling direct estimation of functional connectivity within- and across-scales. Crucially, mPFMs emerged from data-driven multilevel Bayesian modelling of large functional MRI (fMRI) populations. We demonstrate that mPFMs capture both distributed brain modes and their co-existing subcomponents. In addition to validating mPFMs using simulations and real data, we show that mPFMs can predict ~900 personalised traits from UK Biobank more accurately than current standard techniques. Therefore, mPFMs can offer a paradigm shift in functional connectivity modelling and yield enhanced fMRI biomarkers for traits and diseases.

Co-representation of Functional Brain Networks Is Shaped by Cortical Myeloarchitecture and Reveals Individual Behavioral Ability

Large-scale functional networks are spatially distributed in the human brain. Despite recent progress in differentiating their functional roles, how the brain navigates the spatial coordination among them and the biological relevance of this coordination is still not fully understood. Capitalizing on canonical individualized networks derived from functional MRI data, we proposed a new concept, that is, co-representation of functional brain networks, to delineate the spatial coordination among them. To further quantify the co-representation pattern, we defined two indexes, that is, the co-representation specificity (CoRS) and intensity (CoRI), for separately measuring the extent of specific and average expression of functional networks at each brain location by using the data from both sexes. We found that the identified pattern of co-representation was anchored by cortical regions with three types of cytoarchitectural classes along a sensory-fugal axis, including, at the first end, primary (idiotypic) regions showing high CoRS, at the second end, heteromodal regions showing low CoRS and high CoRI, at the third end, paralimbic regions showing low CoRI. Importantly, we demonstrated the critical role of myeloarchitecture in sculpting the spatial distribution of co-representation by assessing the association with the myelin-related neuroanatomical and transcriptomic profiles. Furthermore, the significance of manifesting the co-representation was revealed in its prediction of individual behavioral ability. Our findings indicated that the spatial coordination among functional networks was built upon an anatomically configured blueprint to facilitate neural information processing, while advancing our understanding of the topographical organization of the brain by emphasizing the assembly of functional networks.

Nuclei-specific hypothalamus networks predict a dimensional marker of stress in humans

The hypothalamus is part of the hypothalamic-pituitary-adrenal axis which activates stress responses through release of cortisol. It is a small but heterogeneous structure comprising multiple nuclei. In vivo human neuroimaging has rarely succeeded in recording signals from individual hypothalamus nuclei. Here we use human resting-state fMRI (n = 498) with high spatial resolution to examine relationships between the functional connectivity of specific hypothalamic nuclei and a dimensional marker of prolonged stress. First, we demonstrate that we can parcellate the human hypothalamus into seven nuclei in vivo. Using the functional connectivity between these nuclei and other subcortical structures including the amygdala, we significantly predict stress scores out-of-sample. Predictions use 0.0015% of all possible brain edges, are specific to stress, and improve when using nucleus-specific compared to whole-hypothalamus connectivity. Thus, stress relates to connectivity changes in precise and functionally meaningful subcortical networks, which may be exploited in future studies using interventions in stress disorders.

On the stability of canonical correlation analysis and partial least squares with application to brain-behavior associations

Associations between datasets can be discovered through multivariate methods like Canonical Correlation Analysis (CCA) or Partial Least Squares (PLS). A requisite property for interpretability and generalizability of CCA/PLS associations is stability of their feature patterns. However, stability of CCA/PLS in high-dimensional datasets is questionable, as found in empirical characterizations. To study these issues systematically, we developed a generative modeling framework to simulate synthetic datasets. We found that when sample size is relatively small, but comparable to typical studies, CCA/PLS associations are highly unstable and inaccurate; both in their magnitude and importantly in the feature pattern underlying the association. We confirmed these trends across two neuroimaging modalities and in independent datasets with n ≈ 1000 and n = 20,000, and found that only the latter comprised sufficient observations for stable mappings between imaging-derived and behavioral features. We further developed a power calculator to provide sample sizes required for stability and reliability of multivariate analyses. Collectively, we characterize how to limit detrimental effects of overfitting on CCA/PLS stability, and provide recommendations for future studies.

Graph Learning for Cortical Parcellation from Tensor Decompositions of Resting-State fMRI

Cortical parcellation has long been a cornerstone in the field of neuroscience, enabling the cerebral cortex to be partitioned into distinct, non-overlapping regions that facilitate the interpretation and comparison of complex neuroscientific data. In recent years, these parcellations have frequently been based on the use of resting-state fMRI (rsfMRI) data. In parallel, methods such as independent components analysis have long been used to identify large-scale functional networks with significant spatial overlap between networks. Despite the fact that both forms of decomposition make use of the same spontaneous brain activity measured with rsfMRI, a gap persists in establishing a clear relationship between disjoint cortical parcellations and brain-wide networks. To address this, we introduce a novel parcellation framework that integrates NASCAR, a three-dimensional tensor decomposition method that identifies a series of functional brain networks, with state-of-the-art graph representation learning to produce cortical parcellations that represent near-homogeneous functional regions that are consistent with these brain networks. Further, through the use of the tensor decomposition, we avoid the limitations of traditional approaches that assume statistical independence or orthogonality in defining the underlying networks. Our findings demonstrate that these parcellations are comparable or superior to established atlases in terms of homogeneity of the functional connectivity across parcels, task contrast alignment, and architectonic map alignment. Our methodological pipeline is highly automated, allowing for rapid adaptation to new datasets and the generation of custom parcellations in just minutes, a significant advancement over methods that require extensive manual input. We describe this integrated approach, which we refer to as Untamed, as a tool for use in the fields of cognitive and clinical neuroscientific research. Parcellations created from the Human Connectome Project dataset using Untamed, along with the code to generate atlases with custom parcel numbers, are publicly available at https://untamed-atlas.github.io.

Automating the Human Connectome Project's Temporal ICA Pipeline

Functional magnetic resonance imaging (fMRI) data are dominated by noise and artifacts, with only a small fraction of the variance relating to neural activity. Temporal independent component analysis (tICA) is a recently developed method that enables selective denoising of fMRI artifacts related to physiology such as respiration. However, an automated and easy to use pipeline for tICA has not previously been available; instead, two manual steps have been necessary: 1) setting the group spatial ICA dimensionality after MELODIC's Incremental Group-PCA (MIGP) and 2) labeling tICA components as artifacts versus signals. Moreover, guidance has been lacking as to how many subjects and timepoints are needed to adequately re-estimate the temporal ICA decomposition and what alternatives are available for smaller groups or even individual subjects. Here, we introduce a nine-step fully automated tICA pipeline which removes global artifacts from fMRI dense timeseries after sICA+FIX cleaning and MSMAll alignment driven by functionally relevant areal features. Additionally, we have developed an automated "reclean" Pipeline for improved spatial ICA (sICA) artifact removal. Two major automated components of the pipeline are 1) an automatic group spatial ICA (sICA) dimensionality selection for MIGP data enabled by fitting multiple Wishart distributions; 2) a hierarchical classifier to distinguish group tICA signal components from artifactual components, equipped with a combination of handcrafted features from domain expert knowledge and latent features obtained via self-supervised learning on spatial maps. We demonstrate that the dimensionality estimated for the MIGP data from HCP Young Adult 3T and 7T datasets is comparable to previous manual tICA estimates, and that the group sICA decomposition is highly reproducible. We also show that the tICA classifier achieved over 0.98 Precision-Recall Area Under Curve (PR-AUC) and that the correctly classified components account for over 95% of the tICA-represented variance on multiple held-out evaluation datasets including the HCP-Young Adult, HCP-Aging and HCP-Development datasets under various settings. Our automated tICA pipeline is now available as part of the HCP pipelines, providing a powerful and user-friendly tool for the neuroimaging community.

Data-driven multimodal fusion: approaches and applications in psychiatric research

In the era of big data, where vast amounts of information are being generated and collected at an unprecedented rate, there is a pressing demand for innovative data-driven multi-modal fusion methods. These methods aim to integrate diverse neuroimaging perspectives to extract meaningful insights and attain a more comprehensive understanding of complex psychiatric disorders. However, analyzing each modality separately may only reveal partial insights or miss out on important correlations between different types of data. This is where data-driven multi-modal fusion techniques come into play. By combining information from multiple modalities in a synergistic manner, these methods enable us to uncover hidden patterns and relationships that would otherwise remain unnoticed. In this paper, we present an extensive overview of data-driven multimodal fusion approaches with or without prior information, with specific emphasis on canonical correlation analysis and independent component analysis. The applications of such fusion methods are wide-ranging and allow us to incorporate multiple factors such as genetics, environment, cognition, and treatment outcomes across various brain disorders. After summarizing the diverse neuropsychiatric magnetic resonance imaging fusion applications, we further discuss the emerging neuroimaging analyzing trends in big data, such as N-way multimodal fusion, deep learning approaches, and clinical translation. Overall, multimodal fusion emerges as an imperative approach providing valuable insights into the underlying neural basis of mental disorders, which can uncover subtle abnormalities or potential biomarkers that may benefit targeted treatments and personalized medical interventions.

Anatomically compliant modes of variations: New tools for brain connectivity

Anatomical complexity and data dimensionality present major issues when analysing brain connectivity data. The functional and anatomical aspects of the connections taking place in the brain are in fact equally relevant and strongly intertwined. However, due to theoretical challenges and computational issues, their relationship is often overlooked in neuroscience and clinical research. In this work, we propose to tackle this problem through Smooth Functional Principal Component Analysis, which enables to perform dimensional reduction and exploration of the variability in functional connectivity maps, complying with the formidably complicated anatomy of the grey matter volume. In particular, we analyse a population that includes controls and subjects affected by schizophrenia, starting from fMRI data acquired at rest and during a task-switching paradigm. For both sessions, we first identify the common modes of variation in the entire population. We hence explore whether the subjects' expressions along these common modes of variation differ between controls and pathological subjects. In each session, we find principal components that are significantly differently expressed in the healthy vs pathological subjects (with p-values < 0.001), highlighting clearly interpretable differences in the connectivity in the two subpopulations. For instance, the second and third principal components for the rest session capture the imbalance between the Default Mode and Executive Networks characterizing schizophrenia patients.

A Siamese Network With Node Convolution for Individualized Predictions Based on Connectivity Maps Extracted From Resting-State fMRI Data

Deep learning has demonstrated great potential for objective diagnosis of neuropsychiatric disorders based on neuroimaging data, which includes the promising resting-state functional magnetic resonance imaging (RS-fMRI). However, the insufficient sample size has long been a bottleneck for deep model training for the purpose. In this study, we proposed a Siamese network with node convolution (SNNC) for individualized predictions based on RS-fMRI data. With the involvement of Siamese network, which uses sample pair (rather than a single sample) as input, the problem of insufficient sample size can largely be alleviated. To adapt to connectivity maps extracted from RS-fMRI data, we applied node convolution to each of the two branches of the Siamese network. For regression purposes, we replaced the contrastive loss in classic Siamese network with the mean square error loss and thus enabled Siamese network to quantitatively predict label differences. The label of a test sample can be predicted based on any of the training samples, by adding the label of the training sample to the predicted label difference between them. The final prediction for a test sample in this study was made by averaging the predictions based on each of the training samples. The performance of the proposed SNNC was evaluated with age and IQ predictions based on a public dataset (Cam-CAN). The results indicated that SNNC can make effective predictions even with a sample size of as small as 40, and SNNC achieved state-of-the-art accuracy among a variety of deep models and standard machine learning approaches.

An automated pipeline for obtaining labeled ICA-templates corresponding to functional brain systems

The complexity of our actions and thinking is likely reflected in functional brain networks. Independent component analysis (ICA) is a popular data-driven method to compute group differences between such networks. A common way to investigate network differences is based on ICA maps which are generated from study-specific samples. However, this approach limits the generalizability and reproducibility of the results. Alternatively, network ICA templates can be used, but up to date, few such templates exist and are limited in terms of the functional systems they cover. Here, we propose a simple two-step procedure to obtain ICA-templates corresponding to functional brain systems of the researcher's choice: In step 1, the functional system of interest needs to be defined by means of a statistical parameter map (input), which one can generate with open-source software such as NeuroSynth or BrainMap. In step 2, that map is correlated to group-ICA maps provided by the Human Connectome Project (HCP), which is based on a large sample size and uses high quality and standardized acquisition procedures. The HCP-provided ICA-map with the highest correlation to the input map is then used as an ICA template representing the functional system of interest, for example, for subsequent analyses such as dual regression. We provide a toolbox to complete step 2 of the suggested procedure and demonstrate the usage of our pipeline by producing an ICA templates that corresponds to "motor function" and nine additional brain functional systems resulting in an ICA maps with excellent alignment with the gray matter/white matter boundaries of the brain. Our toolbox generates data in two different file formats: volumetric-based (NIFTI) and combined surface/volumetric files (CIFTI). Compared to 10 existing templates, our procedure output component maps with systematically stronger contribution of gray matter to the ICA z-values compared to white matter voxels in 9/10 cases by at least a factor of 2. The toolbox allows users to investigate functional networks of interest, which will enhance interpretability, reproducibility, and standardization of research investigating functional brain networks.

Resting-State Functional Connectivity Explained Psychotic-like Experiences in the General Population and Partially Generalized to Patients and Relatives

Background

Psychotic-like experiences (PLEs) are considered the subclinical portion of the psychosis continuum. Research suggests that there are resting-state functional connectivity (rsFC) substrates of PLEs, yet it is unclear if the same substrates underlie more severe psychosis. Here, to our knowledge, we report the first study to build a cross-validated rsFC model of PLEs in a large community sample and directly test its ability to explain psychosis in an independent sample of patients with psychosis and their relatives.

Methods

Resting-state FC of 855 healthy young adults from the WU-Minn Human Connectome Project (HCP) was used to predict PLEs with elastic net. An rsFC composite score based on the resulting model was correlated with psychotic traits and symptoms in 118 patients with psychosis, 71 nonpsychotic first-degree relatives, and 45 healthy control subjects from the psychosis HCP.

Results

In the HCP, the cross-validated model explained 3.3% of variance in PLEs. Predictive connections spread primarily across the default, frontoparietal, cingulo-opercular, and dorsal attention networks. The model partially generalized to a younger, but not older, subsample in the psychosis HCP, explaining two measures of positive/disorganized psychotic traits (the Structured Interview for Schizotypy: β = 0.25, pone-tailed = .027; the Schizotypy Personality Questionnaire positive factor: β = 0.14, pone-tailed = .041). However, it did not differentiate patients from relatives and control subjects or explain psychotic symptoms in patients.

Conclusions

Some rsFC substrates of PLEs are shared across the psychosis continuum. However, explanatory power was modest, and generalization was partial. It is equally important to understand shared versus distinct rsFC variances across the psychosis continuum.

Functional brain network identification and fMRI augmentation using a VAE-GAN framework

Recently, deep learning models have achieved superior performance for mapping functional brain networks from functional magnetic resonance imaging (fMRI) data compared with traditional methods. However, due to the lack of sufficient data and the high dimensionality of brain volume, deep learning models of fMRI tend to suffer from overfitting. In addition, existing methods rarely studied fMRI data augmentation and its application. To address these issues, we developed a VAE-GAN framework that combined a VAE (variational auto-encoder) with a GAN (generative adversarial net) for functional brain network identification and fMRI augmentation. As a generative model, the VAE-GAN models the distribution of fMRI so that it enables the extraction of more generalized features, and thus relieve the overfitting issue. The VAE-GAN is easier to train on fMRI than a standard GAN since it uses latent variables from VAE to generate fake data rather than relying on random noise that is used in a GAN, and it can generate higher quality of fake data than VAE since the discriminator can promote the training of the generator. In other words, the VAE-GAN inherits the advantages of VAE and GAN and avoids their limitations in modeling of fMRI data. Extensive experiments on task fMRI datasets from HCP have proved the effectiveness and superiority of the proposed VAE-GAN framework for identifying both temporal features and functional brain networks compared with existing models, and the quality of fake data is higher than those from VAE and GAN. The results on resting state fMRI of Attention Deficit Hyperactivity Disorder (ADHD)-200 dataset further demonstrated that the fake data generated by the VAE-GAN can help improve the performance of brain network modeling and ADHD classification.

A deep learning method for autism spectrum disorder identification based on interactions of hierarchical brain networks

BACKGROUND

It has been recently shown that deep learning models exhibited remarkable performance of representing functional Magnetic Resonance Imaging (fMRI) data for the understanding of brain functional activities. With hierarchical structure, deep learning models can infer hierarchical functional brain networks (FBN) from fMRI. However, the applications of the hierarchical FBNs have been rarely studied.

METHODS

In this work, we proposed a hierarchical recurrent variational auto-encoder (HRVAE) to unsupervisedly model the fMRI data. The trained HRVAE encoder can predict hierarchical temporal features from its three hidden layers, and thus can be regarded as a hierarchical feature extractor. Then LASSO (least absolute shrinkage and selection operator) regression was applied to estimate the corresponding hierarchical FBNs. Based on the hierarchical FBNs from each subject, we constructed a novel classification framework for brain disorder identification and test it on the Autism Brain Imaging Data Exchange (ABIDE) dataset, a world-wide multi-site database of autism spectrum disorder (ASD). We analyzed the hierarchy organization of FBNs, and finally used the overlaps of hierarchical FBNs as features to differentiate ASD from typically developing controls (TDC).

RESULTS

The experimental results on 871 subjects from ABIDE dataset showed that the HRVAE model can effectively derive hierarchical FBNs including many well-known resting state networks (RSN). Moreover, the classification result improved the state-of-the-art by achieving a very high accuracy of 82.1 %.

CONCLUSIONS

This work presents a novel data-driven deep learning method using fMRI data for ASD identification, which could provide valuable reference for clinical diagnosis. The classification results suggest that the interactions of hierarchical FBNs have association with brain disorder, which promotes the understanding of FBN hierarchy and could be applied to other brain disorder analysis.

Generalizing prediction of task-evoked brain activity across datasets and populations

Predictions of task-based functional magnetic resonance imaging (fMRI) from task-free resting-state (rs) fMRI have gained popularity over the past decade. This method holds a great promise for studying individual variability in brain function without the need to perform highly demanding tasks. However, in order to be broadly used, prediction models must prove to generalize beyond the dataset they were trained on. In this work, we test the generalizability of prediction of task-fMRI from rs-fMRI across sites, MRI vendors and age-groups. Moreover, we investigate the data requirements for successful prediction. We use the Human Connectome Project (HCP) dataset to explore how different combinations of training sample sizes and number of fMRI datapoints affect prediction success in various cognitive tasks. We then apply models trained on HCP data to predict brain activations in data from a different site, a different MRI vendor (Phillips vs. Siemens scanners) and a different age group (children from the HCP-development project). We demonstrate that, depending on the task, a training set of approximately 20 participants with 100 fMRI timepoints each yields the largest gain in model performance. Nevertheless, further increasing sample size and number of timepoints results in significantly improved predictions, until reaching approximately 450-600 training participants and 800-1000 timepoints. Overall, the number of fMRI timepoints influences prediction success more than the sample size. We further show that models trained on adequate amounts of data successfully generalize across sites, vendors and age groups and provide predictions that are both accurate and individual-specific. These findings suggest that large-scale publicly available datasets may be utilized to study brain function in smaller, unique samples.

Spatial Smoothing Effect on Group-Level Functional Connectivity during Resting and Task-Based fMRI

Spatial smoothing is a preprocessing step applied to neuroimaging data to enhance data quality by reducing noise and artifacts. However, selecting an appropriate smoothing kernel size can be challenging as it can lead to undesired alterations in final images and functional connectivity networks. However, there is no sufficient information about the effects of the Gaussian kernel size on group-level results for different cases yet. This study investigates the influence of kernel size on functional connectivity networks and network parameters in whole-brain rs-fMRI and tb-fMRI analyses of healthy adults. The analysis includes {0, 2, 4, 6, 8, 10} mm kernels, commonly used in practical analyses, covering all major brain networks. Graph theoretical measures such as betweenness centrality, global/local efficiency, clustering coefficient, and average path length are examined for each kernel. Additionally, principal component analysis (PCA) and independent component analysis (ICA) parameters, namely kurtosis and skewness, are evaluated for the functional images. The findings demonstrate that kernel size directly affects node connections, resulting in modifications to functional network structures and PCA/ICA parameters. However, network metrics exhibit greater resilience to these changes.

Maturational networks of human fetal brain activity reveal emerging connectivity patterns prior to ex-utero exposure

A key feature of the fetal period is the rapid emergence of organised patterns of spontaneous brain activity. However, characterising this process in utero using functional MRI is inherently challenging and requires analytical methods which can capture the constituent developmental transformations. Here, we introduce a novel analytical framework, termed "maturational networks" (matnets), that achieves this by modelling functional networks as an emerging property of the developing brain. Compared to standard network analysis methods that assume consistent patterns of connectivity across development, our method incorporates age-related changes in connectivity directly into network estimation. We test its performance in a large neonatal sample, finding that the matnets approach characterises adult-like features of functional network architecture with a greater specificity than a standard group-ICA approach; for example, our approach is able to identify a nearly complete default mode network. In the in-utero brain, matnets enables us to reveal the richness of emerging functional connections and the hierarchy of their maturational relationships with remarkable anatomical specificity. We show that the associative areas play a central role within prenatal functional architecture, therefore indicating that functional connections of high-level associative areas start emerging prior to exposure to the extra-utero environment.

Structural and functional asymmetry of the neonatal cerebral cortex

Features of brain asymmetry have been implicated in a broad range of cognitive processes; however, their origins are still poorly understood. Here we investigated cortical asymmetries in 442 healthy term-born neonates using structural and functional magnetic resonance images from the Developing Human Connectome Project. Our results demonstrate that the neonatal cortex is markedly asymmetric in both structure and function. Cortical asymmetries observed in the term cohort were contextualized in two ways: by comparing them against cortical asymmetries observed in 103 preterm neonates scanned at term-equivalent age, and by comparing structural asymmetries against those observed in 1,110 healthy young adults from the Human Connectome Project. While associations with preterm birth and biological sex were minimal, significant differences exist between birth and adulthood.

Graph-in-Graph (GiG): Learning interpretable latent graphs in non-Euclidean domain for biological and healthcare applications

Graphs are a powerful tool for representing and analyzing unstructured, non-Euclidean data ubiquitous in the healthcare domain. Two prominent examples are molecule property prediction and brain connectome analysis. Importantly, recent works have shown that considering relationships between input data samples has a positive regularizing effect on the downstream task in healthcare applications. These relationships are naturally modeled by a (possibly unknown) graph structure between input samples. In this work, we propose Graph-in-Graph (GiG), a neural network architecture for protein classification and brain imaging applications that exploits the graph representation of the input data samples and their latent relation. We assume an initially unknown latent-graph structure between graph-valued input data and propose to learn a parametric model for message passing within and across input graph samples, end-to-end along with the latent structure connecting the input graphs. Further, we introduce a Node Degree Distribution Loss (NDDL) that regularizes the predicted latent relationships structure. This regularization can significantly improve the downstream task. Moreover, the obtained latent graph can represent patient population models or networks of molecule clusters, providing a level of interpretability and knowledge discovery in the input domain, which is of particular value in healthcare.

Multi-head attention-based masked sequence model for mapping functional brain networks

The investigation of functional brain networks (FBNs) using task-based functional magnetic resonance imaging (tfMRI) has gained significant attention in the field of neuroimaging. Despite the availability of several methods for constructing FBNs, including traditional methods like GLM and deep learning methods such as spatiotemporal self-attention mechanism (STAAE), these methods have design and training limitations. Specifically, they do not consider the intrinsic characteristics of fMRI data, such as the possibility that the same signal value at different time points could represent different brain states and meanings. Furthermore, they overlook prior knowledge, such as task designs, during training. This study aims to overcome these limitations and develop a more efficient model by drawing inspiration from techniques in the field of natural language processing (NLP). The proposed model, called the Multi-head Attention-based Masked Sequence Model (MAMSM), uses a multi-headed attention mechanism and mask training approach to learn different states corresponding to the same voxel values. Additionally, it combines cosine similarity and task design curves to construct a novel loss function. The MAMSM was applied to seven task state datasets from the Human Connectome Project (HCP) tfMRI dataset. Experimental results showed that the features acquired by the MAMSM model exhibit a Pearson correlation coefficient with the task design curves above 0.95 on average. Moreover, the model can extract more meaningful networks beyond the known task-related brain networks. The experimental results demonstrated that MAMSM has great potential in advancing the understanding of functional brain networks.

Identification of overlapping and interacting networks reveals intrinsic spatiotemporal organization of the human brain

The human brain is a complex network that exhibits dynamic fluctuations in activity across space and time. Depending on the analysis method, canonical brain networks identified from resting-state fMRI (rs-fMRI) are typically constrained to be either orthogonal or statistically independent in their spatial and/or temporal domains. We avoid imposing these potentially unnatural constraints through the combination of a temporal synchronization process ("BrainSync") and a three-way tensor decomposition method ("NASCAR") to jointly analyze rs-fMRI data from multiple subjects. The resulting set of interacting networks comprises minimally constrained spatiotemporal distributions, each representing one component of functionally coherent activity across the brain. We show that these networks can be clustered into six distinct functional categories and naturally form a representative functional network atlas for a healthy population. This functional network atlas could help explore group and individual differences in neurocognitive function, as we demonstrate in the context of ADHD and IQ prediction.

Brain Network Analysis: A Review on Multivariate Analytical Methods

Despite the explosive growth of neuroimaging studies aimed at analyzing the brain as a complex system, critical methodological gaps remain to be addressed. Most tools currently used for analyzing network data of the brain are univariate in nature and are based on assumptions borne out of previous techniques not directly related to the big and complex data of the brain. Although graph-based methods have shown great promise, the development of principled multivariate models to address inherent limitations of graph-based methods, such as their dependence on network size and degree distributions, and to allow assessing the effects of multiple phenotypes on the brain and simulating brain networks has largely lagged behind. Although some studies have been made in developing multivariate frameworks to fill this gap, in the absence of a "gold-standard" method or guidelines, choosing the most appropriate method for each study can be another critical challenge for investigators in this multidisciplinary field. Here, we briefly introduce important multivariate methods for brain network analyses in two main categories: data-driven and model-based methods. We discuss whether/how such methods are suited for examining connectivity (edge-level), topology (system-level), or both. This review will aid in choosing an appropriate multivariate method with respect to variables such as network type, number of subjects and brain regions included, and the interest in connectivity, topology, or both. This review is aimed to be accessible to investigators from different backgrounds, with a focus on applications in brain network studies, though the methods may be applicable in other areas too. Impact statement As the U.S. National Institute of Health notes, the rich biomedical data can greatly improve our knowledge of human health if new analytical tools are developed, and their applications are broadly disseminated. A major challenge in analyzing the brain as a complex system is about developing parsimonious multivariate methods, and particularly choosing the most appropriate one among the existing methods with respect to the study variables in this multidisciplinary field. This study provides a review on the most important multivariate methods to aid in helping the most appropriate ones with respect to the desired variables for each study.

Relationship between nuclei-specific amygdala connectivity and mental health dimensions in humans

There has been increasing interest in using neuroimaging measures to predict psychiatric disorders. However, predictions usually rely on large brain networks and large disorder heterogeneity. Thus, they lack both anatomical and behavioural specificity, preventing the advancement of targeted interventions. Here we address both challenges. First, using resting-state functional magnetic resonance imaging, we parcellated the amygdala, a region implicated in mood disorders, into seven nuclei. Next, a questionnaire factor analysis provided subclinical mental health dimensions frequently altered in anxious-depressive individuals, such as negative emotions and sleep problems. Finally, for each behavioural dimension, we identified the most predictive resting-state functional connectivity between individual amygdala nuclei and highly specific regions of interest, such as the dorsal raphe nucleus in the brainstem or medial frontal cortical regions. Connectivity in circumscribed amygdala networks predicted behaviours in an independent dataset. Our results reveal specific relations between mental health dimensions and connectivity in precise subcortical networks.

Association of Cardiovascular Risk Markers and Fitness with Task-Related Neural Activity during Animacy Perception

PURPOSE

Numerous studies have demonstrated the association between cardiovascular risk markers and fitness, and broad aspects of cognition; however, the possible association of cardiovascular risk markers and fitness with social cognition, which plays a significant role in the development and maintenance of social relationships, has largely been ignored. Herein, we investigated the relationship of cardiovascular risk markers and fitness with task-related neural activity during animacy perception.

METHODS

We analyzed data from the Human Connectome Project derived from 1027 adults age 22-37 yr. Canonical correlation analysis (CCA) was conducted to evaluate the association between participants' body mass index, systolic and diastolic blood pressure, submaximal endurance, gait speed, hand dexterity, and muscular strength with task-related neural activity during animacy perception.

RESULTS

We observed a single significant CCA mode. Body mass index and blood pressure demonstrated negative cross-loadings with task-related neural activity in the temporoparietal, superior and anterior temporal, posterior cingulate, and inferior frontal regions, whereas submaximal endurance, hand dexterity, and muscular strength demonstrated positive cross-loadings. The observed CCA variates did not seem highly heritable, as the absolute differences in CCA variates in monozygotic twins, dizygotic twins, and nontwin siblings were not statistically different. Furthermore, the cardiovascular risk markers and fitness CCA variates were positively associated with animacy perception and emotion recognition accuracy, which was mediated by the task-related neural activity.

CONCLUSIONS

The present findings can provide new insights into the role of markers for cardiovascular health and fitness, specifically their association with social cognition and the underlying neural basis. The intervention for cardiovascular risk and fitness could be a potentially cost-effective method of targeting social cognition.

Accurate predictions of individual differences in task-evoked brain activity from resting-state fMRI using a sparse ensemble learner

Modelling and predicting individual differences in task-fMRI activity can have a wide range of applications from basic to clinical neuroscience. It has been shown that models based on resting-state activity can have high predictive accuracy. Here we propose several improvements to such models. Using a sparse ensemble learner, we show that (i) features extracted using Stochastic Probabilistic Functional Modes (sPROFUMO) outperform the previously proposed dual-regression approach, (ii) that the shape and overall intensity of individualised task activations can be modelled separately and explicitly, (iii) training the model on predicting residual differences in brain activity further boosts individualised predictions. These results hold for both surface-based analyses of the Human Connectome Project data as well as volumetric analyses of UK-biobank data. Overall, our model achieves state of the art prediction accuracy on par with the test-retest reliability of task-fMRI scans, suggesting that it has potential to supplement traditional task localisers.

Active neural coordination of motor behaviors with internal states

The brain continuously coordinates skeletomuscular movements with internal physiological states like arousal, but how is this coordination achieved? One possibility is that the brain simply reacts to changes in external and/or internal signals. Another possibility is that it is actively coordinating both external and internal activities. We used functional ultrasound imaging to capture a large medial section of the brain, including multiple cortical and subcortical areas, in marmoset monkeys while monitoring their spontaneous movements and cardiac activity. By analyzing the causal ordering of these different time series, we found that information flowing from the brain to movements and heart-rate fluctuations were significantly greater than in the opposite direction. The brain areas involved in this external versus internal coordination were spatially distinct, but also extensively interconnected. Temporally, the brain alternated between network states for this regulation. These findings suggest that the brain's dynamics actively and efficiently coordinate motor behavior with internal physiology.

Using deep clustering to improve fMRI dynamic functional connectivity analysis

Dynamic functional connectivity (dFC) analysis of resting-state fMRI data is commonly performed by calculating sliding-window correlations (SWC), followed by k-means clustering in order to assign each window to a given state. Studies using synthetic data have shown that k-means performance is highly dependent on sliding window parameters and signal-to-noise ratio. Additionally, sources of heterogeneity between subjects may affect the accuracy of group-level clustering, thus affecting measurements of dFC state temporal properties such as dwell time and fractional occupancy. This may result in spurious conclusions regarding differences between groups (e.g. when comparing a clinical population to healthy controls). Therefore, is it important to quantify the ability of k-means to estimate dFC state temporal properties when applied to cohorts of multiple subjects, and to explore ways in which clustering performance can be maximised. Here, we explore the use of dimensionality reduction methods prior to clustering in order to map high-dimensional data to a lower dimensional space, providing salient features to the subsequent clustering step. We assess the use of deep autoencoders for dimensionality reduction prior to applying k-means clustering to the encoded data. We compare this deep clustering method to dimensionality reduction using principle component analysis (PCA), uniform manifold approximation and projection (UMAP), as well as applying k-means to the original feature space using either L1 or L2 distance. We provide extensive quantitative evaluation of clustering performance using synthetic datasets, representing data from multiple heterogeneous subjects. In synthetic data we find that deep clustering gives the best performance, while other approaches are often insufficient to capture temporal properties of dFC states. We then demonstrate the application of each method to real-world data from human subjects and show that the choice of dimensionality reduction method has a significant effect on group-level measurements of state temporal properties.

Learning brain representation using recurrent Wasserstein generative adversarial net

BACKGROUND AND OBJECTIVE

To understand brain cognition and disorders, modeling the mapping between mind and brain has been of great interest to the neuroscience community. The key is the brain representation, including functional brain networks (FBN) and their corresponding temporal features. Recently, it has been proven that deep learning models have superb representation power on functional magnetic resonance imaging (fMRI) over traditional machine learning methods. However, due to the lack of high-quality data and labels, deep learning models tend to suffer from overfitting in the training process.

METHODS

In this work, we applied a recurrent Wasserstein generative adversarial net (RWGAN) to learn brain representation from volumetric fMRI data. Generative adversarial net (GAN) is widely used in natural image generation and is able to capture the distribution of the input data, which enables the extraction of generalized features from fMRI and thus relieves the overfitting issue. The recurrent layers in RWGAN are designed to better model the local temporal features of the fMRI time series. The discriminator of RWGAN works as a deep feature extractor. With LASSO regression, the RWGAN model can decompose the fMRI data into temporal features and spatial features (FBNs). Furthermore, the generator of RWGAN can generate high-quality new data for fMRI augmentation.

RESULTS

The experimental results on seven tasks from the HCP dataset showed that the RWGAN can learn meaningful and interpretable temporal features and FBNs, compared to HCP task designs and general linear model (GLM) derived networks. Besides, the results on different training datasets showed that the RWGAN performed better on small datasets than other deep learning models. Moreover, we used the generator of RWGAN to yield fake subjects. The result showed that the fake data can also be used to learn meaningful representation compared to those learned from real data.

CONCLUSIONS

To our best knowledge, this work is among the earliest attempts of applying generative deep learning for modeling fMRI data. The proposed RWGAN offers a novel methodology for learning brain representation from fMRI, and it can generate high-quality fake data for the potential use of fMRI data augmentation.

Dynamical differential covariance recovers directional network structure in multiscale neural systems

Investigating neural interactions is essential to understanding the neural basis of behavior. Many statistical methods have been used for analyzing neural activity, but estimating the direction of network interactions correctly and efficiently remains a difficult problem. Here, we derive dynamical differential covariance (DDC), a method based on dynamical network models that detects directional interactions with low bias and high noise tolerance under nonstationarity conditions. Moreover, DDC scales well with the number of recording sites and the computation required is comparable to that needed for covariance. DDC was validated and compared favorably with other methods on networks with false positive motifs and multiscale neural simulations where the ground-truth connectivity was known. When applied to recordings of resting-state functional magnetic resonance imaging (rs-fMRI), DDC consistently detected regional interactions with strong structural connectivity in over 1,000 individual subjects obtained by diffusion MRI (dMRI). DDC is a promising family of methods for estimating connectivity that can be generalized to a wide range of dynamical models and recording techniques and to other applications where system identification is needed.

Act natural: Functional connectivity from naturalistic stimuli fMRI outperforms resting-state in predicting brain activity

The search for an 'ideal' approach to investigate the functional connections in the human brain is an ongoing challenge for the neuroscience community. While resting-state functional magnetic resonance imaging (fMRI) has been widely used to study individual functional connectivity patterns, recent work has highlighted the benefits of collecting functional connectivity data while participants are exposed to naturalistic stimuli, such as watching a movie or listening to a story. For example, functional connectivity data collected during movie-watching were shown to predict cognitive and emotional scores more accurately than resting-state-derived functional connectivity. We have previously reported a tight link between resting-state functional connectivity and task-derived neural activity, such that the former successfully predicts the latter. In the current work we use data from the Human Connectome Project to demonstrate that naturalistic-stimulus-derived functional connectivity predicts task-induced brain activation maps more accurately than resting-state-derived functional connectivity. We then show that activation maps predicted using naturalistic stimuli are better predictors of individual intelligence scores than activation maps predicted using resting-state. We additionally examine the influence of naturalistic-stimulus type on prediction accuracy. Our findings emphasize the potential of naturalistic stimuli as a promising alternative to resting-state fMRI for connectome-based predictive modelling of individual brain activity and cognitive traits.

Evaluating brain parcellations using the distance-controlled boundary coefficient

One important approach to human brain mapping is to define a set of distinct regions that can be linked to unique functions. Numerous brain parcellations have been proposed, using cytoarchitectonic, structural, or functional magnetic resonance imaging (fMRI) data. The intrinsic smoothness of brain data, however, poses a problem for current methods seeking to compare different parcellations. For example, criteria that simply compare within‐parcel to between‐parcel similarity provide even random parcellations with a high value. Furthermore, the evaluation is biased by the spatial scale of the parcellation. To address this problem, we propose the distance‐controlled boundary coefficient (DCBC), an unbiased criterion to evaluate discrete parcellations. We employ this new criterion to evaluate existing parcellations of the human neocortex in their power to predict functional boundaries for an fMRI data set with many different tasks, as well as for resting‐state data. We find that common anatomical parcellations do not perform better than chance, suggesting that task‐based functional boundaries do not align well with sulcal landmarks. Parcellations based on resting‐state fMRI data perform well; in some cases, as well as a parcellation defined on the evaluation data itself. Finally, multi‐modal parcellations that combine functional and anatomical criteria perform substantially worse than those based on functional data alone, indicating that functionally homogeneous regions often span major anatomical landmarks. Overall, the DCBC advances the field of functional brain mapping by providing an unbiased metric that compares the predictive ability of different brain parcellations to define brain regions that are functionally maximally distinct.

Multi-dynamic modelling reveals strongly time-varying resting fMRI correlations

The activity of functional brain networks is responsible for the emergence of time-varying cognition and behaviour. Accordingly, time-varying correlations (Functional Connectivity) in resting fMRI have been shown to be predictive of behavioural traits, and psychiatric and neurological conditions. Typically, methods that measure time varying Functional Connectivity (FC), such as sliding windows approaches, do not separately model when changes occur in the mean activity levels from when changes occur in the FC, therefore conflating these two distinct types of modulation. We show that this can bias the estimation of time-varying FC to appear more stable over time than it actually is. Here, we propose an alternative approach that models changes in the mean brain activity and in the FC as being able to occur at different times to each other. We refer to this method as the Multi-dynamic Adversarial Generator Encoder (MAGE) model, which includes a model of the network dynamics that captures long-range time dependencies, and is estimated on fMRI data using principles of Generative Adversarial Networks. We evaluated the approach across several simulation studies and resting fMRI data from the Human Connectome Project (1003 subjects), as well as from UK Biobank (13301 subjects). Importantly, we find that separating fluctuations in the mean activity levels from those in the FC reveals much stronger changes in FC over time, and is a better predictor of individual behavioural variability.

A mixed-modeling framework for whole-brain dynamic network analysis

The emerging area of dynamic brain network analysis has gained considerable attention in recent years. However, development of multivariate statistical frameworks that allow for examining the associations between phenotypic traits and dynamic patterns of system-level properties of the brain, and drawing statistical inference about such associations, has largely lagged behind. To address this need we developed a mixed-modeling framework that allows for assessing the relationship between any desired phenotype and dynamic patterns of whole-brain connectivity and topology. This novel framework also allows for simulating dynamic brain networks with respect to desired covariates. Unlike current tools, which largely use data-driven methods, our model-based method enables aligning neuroscientific hypotheses with the analytic approach. We demonstrate the utility of this model in identifying the relationship between fluid intelligence and dynamic brain networks by using resting-state fMRI (rfMRI) data from 200 participants in the Human Connectome Project (HCP) study. We also demonstrate the utility of this model to simulate dynamic brain networks at both group and individual levels. To our knowledge, this approach provides the first model-based statistical method for examining dynamic patterns of system-level properties of the brain and their relationships to phenotypic traits as well as simulating dynamic brain networks.

In recent years, a growing body of studies have aimed at analyzing the brain as a complex dynamic system by using various neuroimaging data. This has opened new avenues to answer compelling questions about the brain function in health and disease. However, methods that allow for providing statistical inference about how the complex interactions of the brain are associated with desired phenotypes are to be developed for a more profound insight. This study introduces a promising regression-based model to relate dynamic brain networks to desired phenotypes and provide statistical inference. Moreover, it can be used for simulating dynamic brain networks with respect to desired phenotypes at the group and individual levels.

Functional connectivity of fMRI using differential covariance predicts structural connectivity and behavioral reaction times

Recordings from resting state functional Magnetic Resonance Imaging (rs-fMRI) reflect the influence of pathways between brain areas. A wide range of methods have been proposed to measure this functional connectivity (FC), but the lack of “ground truth” has made it difficult to systematically validate them. Most measures of FC produce connectivity estimates that are symmetrical between brain areas. Differential covariance (dCov) is an algorithm for analyzing FC with directed graph edges. When we applied dCov to rs-fMRI recordings from the human connectome project (HCP) and anesthetized mice, dCov-FC accurately identified strong cortical connections from diffusion Magnetic Resonance Imaging (dMRI) in individual humans and viral tract tracing in mice. In addition, those HCP subjects whose dCov-FCs were more integrated, as assessed by a graph-theoretic measure, tended to have shorter reaction times in several behavioral tests. Thus, dCov-FC was able to identify anatomically verified connectivity that yielded measures of brain integration significantly correlated with behavior.

Medial temporal lobe contributions to resting-state networks

The medial temporal lobe (MTL) is a set of interconnected brain regions that have been shown to play a central role in behavior as well as in neurological disease. Recent studies using resting-state functional magnetic resonance imaging (rsfMRI) have attempted to understand the MTL in terms of its functional connectivity with the rest of the brain. However, the exact characterization of the whole-brain networks that co-activate with the MTL as well as how the various sub-regions of the MTL are associated with these networks remains poorly understood. Here, we attempted to advance these issues by exploiting the high spatial resolution 7T rsfMRI dataset from the Human Connectome Project with a data-driven analysis approach that relied on independent component analysis (ICA) restricted to the MTL. We found that four different well-known resting-state networks co-activated with a unique configuration of MTL subcomponents. Specifically, we found that different sections of the parahippocampal cortex were involved in the default mode, visual and dorsal attention networks; sections of the hippocampus in the somatomotor and default mode networks; and the lateral entorhinal cortex in the dorsal attention network. We replicated this set of results in a validation sample. These results provide new insight into how the MTL and its subcomponents contribute to known resting-state networks. The participation of the MTL in an expanded range of resting-state networks is in line with recent proposals on MTL function.

Predicting individual traits from unperformed tasks

Relating individual differences in cognitive traits to brain functional organization is a long-lasting challenge for the neuroscience community. Individual intelligence scores were previously predicted from whole-brain connectivity patterns, extracted from functional magnetic resonance imaging (fMRI) data acquired at rest. Recently, it was shown that task-induced brain activation maps outperform these resting-state connectivity patterns in predicting individual intelligence, suggesting that a cognitively demanding environment improves prediction of cognitive abilities. Here, we use data from the Human Connectome Project to predict task-induced brain activation maps from resting-state fMRI, and proceed to use these predicted activity maps to further predict individual differences in a variety of traits. While models based on original task activation maps remain the most accurate, models based on predicted maps significantly outperformed those based on the resting-state connectome. Thus, we provide a promising approach for the evaluation of measures of human behavior from brain activation maps, that could be used without having participants actually perform the tasks.

Learning Cortical Parcellations Using Graph Neural Networks

Deep learning has been applied to magnetic resonance imaging (MRI) for a variety of purposes, ranging from the acceleration of image acquisition and image denoising to tissue segmentation and disease diagnosis. Convolutional neural networks have been particularly useful for analyzing MRI data due to the regularly sampled spatial and temporal nature of the data. However, advances in the field of brain imaging have led to network- and surface-based analyses that are often better represented in the graph domain. In this analysis, we propose a general purpose cortical segmentation method that, given resting-state connectivity features readily computed during conventional MRI pre-processing and a set of corresponding training labels, can generate cortical parcellations for new MRI data. We applied recent advances in the field of graph neural networks to the problem of cortical surface segmentation, using resting-state connectivity to learn discrete maps of the human neocortex. We found that graph neural networks accurately learn low-dimensional representations of functional brain connectivity that can be naturally extended to map the cortices of new datasets. After optimizing over algorithm type, network architecture, and training features, our approach yielded mean classification accuracies of 79.91% relative to a previously published parcellation. We describe how some hyperparameter choices including training and testing data duration, network architecture, and algorithm choice affect model performance.

Echo state network models for nonlinear Granger causality

While Granger causality (GC) has been often employed in network neuroscience, most GC applications are based on linear multivariate autoregressive (MVAR) models. However, real-life systems like biological networks exhibit notable nonlinear behaviour, hence undermining the validity of MVAR-based GC (MVAR-GC). Most nonlinear GC estimators only cater for additive nonlinearities or, alternatively, are based on recurrent neural networks or long short-term memory networks, which present considerable training difficulties and tailoring needs. We reformulate the GC framework in terms of echo-state networks-based models for arbitrarily complex networks, and characterize its ability to capture nonlinear causal relations in a network of noisy Duffing oscillators, showing a net advantage of echo state GC (ES-GC) in detecting nonlinear, causal links. We then explore the structure of ES-GC networks in the human brain employing functional MRI data from 1003 healthy subjects drawn from the human connectome project, demonstrating the existence of previously unknown directed within-brain interactions. In addition, we examine joint brain-heart signals in 15 subjects where we explore directed interaction between brain networks and central vagal cardiac control in order to investigate the so-called central autonomic network in a causal manner. This article is part of the theme issue 'Advanced computation in cardiovascular physiology: new challenges and opportunities'.

An efficient functional magnetic resonance imaging data reduction strategy using neighborhood preserving embedding algorithm

High dimensionality data have become common in neuroimaging fields, especially group‐level functional magnetic resonance imaging (fMRI) datasets. fMRI connectivity analysis is a widely used, powerful technique for studying functional brain networks to probe underlying mechanisms of brain function and neuropsychological disorders. However, data‐driven technique like independent components analysis (ICA), can yield unstable and inconsistent results, confounding the true effects of interest and hindering the understanding of brain functionality and connectivity. A key contributing factor to this instability is the information loss that occurs during fMRI data reduction. Data reduction of high dimensionality fMRI data in the temporal domain to identify the important information within group datasets is necessary for such analyses and is crucial to ensure the accuracy and stability of the outputs. In this study, we describe an fMRI data reduction strategy based on an adapted neighborhood preserving embedding (NPE) algorithm. Both simulated and real data results indicate that, compared with the widely used data reduction method, principal component analysis, the NPE‐based data reduction method (a) shows superior performance on efficient data reduction, while enhancing group‐level information, (b) develops a unique stratagem for selecting components based on an adjacency graph of eigenvectors, (c) generates more reliable and reproducible brain networks under different model orders when the outputs of NPE are used for ICA, (d) is more sensitive to revealing task‐evoked activation for task fMRI, and (e) is extremely attractive and powerful for the increasingly popular fast fMRI and very large datasets.

Dimensionality Reduction: Foundations and Applications in Clinical Neuroscience

Advancements in population neuroscience are spurred by the availability of large scale, open datasets, such as the Human Connectome Project or recently introduced UK Biobank. With the increasing data availability, analyses of brain imaging data employ more and more sophisticated machine learning algorithms. However, all machine learning algorithms must balance generalization and complexity. As the detail of neuroimaging data leads to high-dimensional data spaces, model complexity and hence the chance of overfitting increases. Different methodological approaches can be applied to alleviate the problems that arise in high-dimensional settings by reducing the original information into meaningful and concise features. One popular approach is dimensionality reduction, which allows to summarize high-dimensional data into low-dimensional representations while retaining relevant trends and patterns. In this paper, principal component analysis (PCA) is discussed as widely used dimensionality reduction method based on current examples of population-based neuroimaging analyses.

Lesion covariance networks reveal proposed origins and pathways of diffuse gliomas

Diffuse gliomas have been hypothesized to originate from neural stem cells in the subventricular zone and develop along previously healthy brain networks. Here, we evaluated these hypotheses by mapping independent sources of glioma localization and determining their relationships with neurogenic niches, genetic markers and large-scale connectivity networks. By applying independent component analysis to lesion data from 242 adult patients with high- and low-grade glioma, we identified three lesion covariance networks, which reflect clusters of frequent glioma localization. Replicability of the lesion covariance networks was assessed in an independent sample of 168 glioma patients. We related the lesion covariance networks to important clinical variables, including tumour grade and patient survival, as well as genomic information such as molecular genetic subtype and bulk transcriptomic profiles. Finally, we systematically cross-correlated the lesion covariance networks with structural and functional connectivity networks derived from neuroimaging data of over 4000 healthy UK BioBank participants to uncover intrinsic brain networks that may that underlie tumour development. The three lesion covariance networks overlapped with the anterior, posterior and inferior horns of the lateral ventricles respectively, extending into the frontal, parietal and temporal cortices. These locations were independently replicated. The first lesion covariance network, which overlapped with the anterior horn, was associated with low-grade, isocitrate dehydrogenase -mutated/1p19q-codeleted tumours, as well as a neural transcriptomic signature and improved overall survival. Each lesion covariance network significantly coincided with multiple structural and functional connectivity networks, with the first bearing an especially strong relationship with brain connectivity, consistent with its neural transcriptomic profile. Finally, we identified subcortical, periventricular structures with functional connectivity patterns to the cortex that significantly matched each lesion covariance network. In conclusion, we demonstrated replicable patterns of glioma localization with clinical relevance and spatial correspondence with large-scale functional and structural connectivity networks. These results are consistent with prior reports of glioma growth along white matter pathways, as well as evidence for the coordination of glioma stem cell proliferation by neuronal activity. Our findings describe how the locations of gliomas relate to their proposed subventricular origins, suggesting a model wherein periventricular brain connectivity guides tumour development.