Neuroimaging of the Philadelphia neurodevelopmental cohort

Development and Maturation of the Human Brain, from Infancy to Adolescence

This chapter describes basic principles and key findings regarding the development and maturation of the human brain, the former referring to the pre-natal and early post-natal periods and the latter concerning childhood and adolescence. In both cases, we focus on brain structure as revealed in vivo with multi-modal magnetic resonance imaging (MRI). We begin with a few numbers about the human brain and its cellular composition and a brief overview of a number of MRI-based metrics used to characterize age-related variations in grey and white matter. We then proceed with synthesizing current knowledge about developmental and maturational changes in the cerebral cortex (its thickness, surface area, and intra-cortical myelination) and the underlying white matter (volume and structural properties). To facilitate biological interpretations of MRI-derived metrics, we introduce the concept of virtual histology. We conclude the chapter with a few notes about future directions in the study of factors shaping the human brain from conception onwards.

Regression and alignment for functional data and network topology

In the brain, functional connections form a network whose topological organization can be described by graph-theoretic network diagnostics. These include characterizations of the community structure, such as modularity and participation coefficient, which have been shown to change over the course of childhood and adolescence. To investigate if such changes in the functional network are associated with changes in cognitive performance during development, network studies often rely on an arbitrary choice of preprocessing parameters, in particular the proportional threshold of network edges. Because the choice of parameter can impact the value of the network diagnostic, and therefore downstream conclusions, we propose to circumvent that choice by conceptualizing the network diagnostic as a function of the parameter. As opposed to a single value, a network diagnostic curve describes the connectome topology at multiple scales-from the sparsest group of the strongest edges to the entire edge set. To relate these curves to executive function and other covariates, we use scalar-on-function regression, which is more flexible than previous functional data-based models used in network neuroscience. We then consider how systematic differences between networks can manifest in misalignment of diagnostic curves, and consequently propose a supervised curve alignment method that incorporates auxiliary information from other variables. Our algorithm performs both functional regression and alignment via an iterative, penalized, and nonlinear likelihood optimization. The illustrated method has the potential to improve the interpretability and generalizability of neuroscience studies where the goal is to study heterogeneity among a mixture of function- and scalar-valued measures.

Explainable spatio-temporal graph evolution learning with applications to dynamic brain network analysis during development

Modeling dynamic interactions among network components is crucial to uncovering the evolution mechanisms of complex networks. Recently, spatio-temporal graph learning methods have achieved noteworthy results in characterizing the dynamic changes of inter-node relations (INRs). However, challenges remain: The spatial neighborhood of an INR is underexploited, and the spatio-temporal dependencies in INRs' dynamic changes are overlooked, ignoring the influence of historical states and local information. In addition, the model's explainability has been understudied. To address these issues, we propose an explainable spatio-temporal graph evolution learning (ESTGEL) model to model the dynamic evolution of INRs. Specifically, an edge attention module is proposed to utilize the spatial neighborhood of an INR at multi-level, i.e., a hierarchy of nested subgraphs derived from decomposing the initial node-relation graph. Subsequently, a dynamic relation learning module is proposed to capture the spatio-temporal dependencies of INRs. The INRs are then used as adjacent information to improve the node representation, resulting in comprehensive delineation of dynamic evolution of the network. Finally, the approach is validated with real data on brain development study. Experimental results on dynamic brain networks analysis reveal that brain functional networks transition from dispersed to more convergent and modular structures throughout development. Significant changes are observed in the dynamic functional connectivity (dFC) associated with functions including emotional control, decision-making, and language processing.

The role of functional emotion circuits in distinct dimensions of psychopathology in youth

Several mental disorders emerge during childhood or adolescence and are often characterized by socioemotional difficulties, including alterations in emotion perception. Emotional facial expressions are processed in discrete functional brain modules whose connectivity patterns encode emotion categories, but the involvement of these neural circuits in psychopathology in youth is poorly understood. This study examined the associations between activation and functional connectivity patterns in emotion circuits and psychopathology during development. We used task-based fMRI data from the Philadelphia Neurodevelopmental Cohort (PNC, N = 1221, 8-23 years) and conducted generalized psycho-physiological interaction (gPPI) analyses. Measures of psychopathology were derived from an independent component analysis of questionnaire data. The results showed positive associations between identifying fearful, sad, and angry faces and depressive symptoms, and a negative relationship between sadness recognition and positive psychosis symptoms. We found a positive main effect of depressive symptoms on BOLD activation in regions overlapping with the default mode network, while individuals reporting higher levels of norm-violating behavior exhibited emotion-specific lower functional connectivity within regions of the salience network and between modules that overlapped with the salience and default mode network. Our findings illustrate the relevance of functional connectivity patterns underlying emotion processing for behavioral problems in children and adolescents.

Large-scale analysis of structural brain asymmetries during neurodevelopment: Associations with age and sex in 4265 children and adolescents

Only a small number of studies have assessed structural differences between the two hemispheres during childhood and adolescence. However, the existing findings lack consistency or are restricted to a particular brain region, a specific brain feature, or a relatively narrow age range. Here, we investigated associations between brain asymmetry and age as well as sex in one of the largest pediatric samples to date (n = 4265), aged 1-18 years, scanned at 69 sites participating in the ENIGMA (Enhancing NeuroImaging Genetics through Meta-Analysis) consortium. Our study revealed that significant brain asymmetries already exist in childhood, but their magnitude and direction depend on the brain region examined and the morphometric measurement used (cortical volume or thickness, regional surface area, or subcortical volume). With respect to effects of age, some asymmetries became weaker over time while others became stronger; sometimes they even reversed direction. With respect to sex differences, the total number of regions exhibiting significant asymmetries was larger in females than in males, while the total number of measurements indicating significant asymmetries was larger in males (as we obtained more than one measurement per cortical region). The magnitude of the significant asymmetries was also greater in males. However, effect sizes for both age effects and sex differences were small. Taken together, these findings suggest that cerebral asymmetries are an inherent organizational pattern of the brain that manifests early in life. Overall, brain asymmetry appears to be relatively stable throughout childhood and adolescence, with some differential effects in males and females.

Power and reproducibility in the external validation of brain-phenotype predictions

Brain-phenotype predictive models seek to identify reproducible and generalizable brain-phenotype associations. External validation, or the evaluation of a model in external datasets, is the gold standard in evaluating the generalizability of models in neuroimaging. Unlike typical studies, external validation involves two sample sizes: the training and the external sample sizes. Thus, traditional power calculations may not be appropriate. Here we ran over 900 million resampling-based simulations in functional and structural connectivity data to investigate the relationship between training sample size, external sample size, phenotype effect size, theoretical power and simulated power. Our analysis included a wide range of datasets: the Healthy Brain Network, the Adolescent Brain Cognitive Development Study, the Human Connectome Project (Development and Young Adult), the Philadelphia Neurodevelopmental Cohort, the Queensland Twin Adolescent Brain Project, and the Chinese Human Connectome Project; and phenotypes: age, body mass index, matrix reasoning, working memory, attention problems, anxiety/depression symptoms and relational processing. High effect size predictions achieved adequate power with training and external sample sizes of a few hundred individuals, whereas low and medium effect size predictions required hundreds to thousands of training and external samples. In addition, most previous external validation studies used sample sizes prone to low power, and theoretical power curves should be adjusted for the training sample size. Furthermore, model performance in internal validation often informed subsequent external validation performance (Pearson's r difference <0.2), particularly for well-harmonized datasets. These results could help decide how to power future external validation studies.

A network control theory pipeline for studying the dynamics of the structural connectome

Network control theory (NCT) is a simple and powerful tool for studying how network topology informs and constrains the dynamics of a system. Compared to other structure-function coupling approaches, the strength of NCT lies in its capacity to predict the patterns of external control signals that may alter the dynamics of a system in a desired way. An interesting development for NCT in the neuroscience field is its application to study behavior and mental health symptoms. To date, NCT has been validated to study different aspects of the human structural connectome. NCT outputs can be monitored throughout developmental stages to study the effects of connectome topology on neural dynamics and, separately, to test the coherence of empirical datasets with brain function and stimulation. Here, we provide a comprehensive pipeline for applying NCT to structural connectomes by following two procedures. The main procedure focuses on computing the control energy associated with the transitions between specific neural activity states. The second procedure focuses on computing average controllability, which indexes nodes' general capacity to control the dynamics of the system. We provide recommendations for comparing NCT outputs against null network models, and we further support this approach with a Python-based software package called 'network control theory for python'. The procedures in this protocol are appropriate for users with a background in network neuroscience and experience in dynamical systems theory.

Adolescent maturation of cortical excitation-inhibition balance based on individualized biophysical network modeling

The balance of excitation and inhibition is a key functional property of cortical microcircuits which changes through the lifespan. Adolescence is considered a crucial period for the maturation of excitation-inhibition balance. This has been primarily observed in animal studies, yet human in vivo evidence on adolescent maturation of the excitation-inhibition balance at the individual level is limited. Here, we developed an individualized in vivo marker of regional excitation-inhibition balance in human adolescents, estimated using large-scale simulations of biophysical network models fitted to resting-state functional magnetic resonance imaging data from two independent cross-sectional (N = 752) and longitudinal (N = 149) cohorts. We found a widespread relative increase of inhibition in association cortices paralleled by a relative age-related increase of excitation, or lack of change, in sensorimotor areas across both datasets. This developmental pattern co-aligned with multiscale markers of sensorimotor-association differentiation. The spatial pattern of excitation-inhibition development in adolescence was robust to inter-individual variability of structural connectomes and modeling configurations. Notably, we found that alternative simulation-based markers of excitation-inhibition balance show a variable sensitivity to maturational change. Taken together, our study highlights an increase of inhibition during adolescence in association areas using cross sectional and longitudinal data, and provides a robust computational framework to estimate microcircuit maturation in vivo at the individual level.

The General Psychopathology 'p' Factor in Adolescence: Multi-Informant Assessment and Computerized Adaptive Testing

Accumulating evidence supports the presence of a general psychopathology dimension, the p factor ('p'). Despite growing interest in the p factor, questions remain about how p is assessed. Although multi-informant assessment of psychopathology is commonplace in clinical research and practice with children and adolescents, almost no research has taken a multi-informant approach to studying youth p or has examined the degree of concordance between parent and youth reports. Further, estimating p requires assessment of a large number of symptoms, resulting in high reporter burden that may not be feasible in many clinical and research settings. In the present study, we used bifactor multidimensional item response theory models to estimate parent- and adolescent-reported p in a large community sample of youth (11-17 years) and parents (N = 5,060 dyads). We examined agreement between parent and youth p scores and associations with assessor-rated youth global functioning. We also applied computerized adaptive testing (CAT) simulations to parent and youth reports to determine whether adaptive testing substantially alters agreement on p or associations with youth global functioning. Parent-youth agreement on p was moderate (r =.44) and both reports were negatively associated with youth global functioning. Notably, 7 out of 10 of the highest loading items were common across reporters. CAT reduced the average number of items administered by 57%. Agreement between CAT-derived p scores was similar to the full form (r =.40) and CAT scores were negatively correlated with youth functioning. These novel results highlight the promise and potential clinical utility of a multi-informant p factor approach.

In vivo whole-cortex marker of excitation-inhibition ratio indexes cortical maturation and cognitive ability in youth

A balanced excitation-inhibition ratio (E/I ratio) is critical for healthy brain function. Normative development of cortex-wide E/I ratio remains unknown. Here, we noninvasively estimate a putative marker of whole-cortex E/I ratio by fitting a large-scale biophysically plausible circuit model to resting-state functional MRI (fMRI) data. We first confirm that our model generates realistic brain dynamics in the Human Connectome Project. Next, we show that the estimated E/I ratio marker is sensitive to the gamma-aminobutyric acid (GABA) agonist benzodiazepine alprazolam during fMRI. Alprazolam-induced E/I changes are spatially consistent with positron emission tomography measurement of benzodiazepine receptor density. We then investigate the relationship between the E/I ratio marker and neurodevelopment. We find that the E/I ratio marker declines heterogeneously across the cerebral cortex during youth, with the greatest reduction occurring in sensorimotor systems relative to association systems. Importantly, among children with the same chronological age, a lower E/I ratio marker (especially in the association cortex) is linked to better cognitive performance. This result is replicated across North American (8.2 to 23.0 y old) and Asian (7.2 to 7.9 y old) cohorts, suggesting that a more mature E/I ratio indexes improved cognition during normative development. Overall, our findings open the door to studying how disrupted E/I trajectories may lead to cognitive dysfunction in psychopathology that emerges during youth.

Domain-specific associations between psychopathology and neurocognitive functioning

BACKGROUND

Neurocognitive dysfunction is a transdiagnostic finding in psychopathology, but relationships among cognitive domains and general and specific psychopathology dimensions remain unclear. This study aimed to examine associations between cognition and psychopathology dimensions in a large youth cohort.

METHOD

The sample (N = 9350; age 8-21 years) was drawn from the Philadelphia Neurodevelopmental Cohort. Data from structured clinical interviews were modeled using bifactor confirmatory factor analysis (CFA), resulting in an overall psychopathology ('p') factor score and six orthogonal psychopathology dimensions: dysphoria/distress, obsessive-compulsive, behavioral/externalizing, attention-deficit/hyperactivity, phobias, and psychosis. Neurocognitive data were aggregated using correlated-traits CFA into five factors: executive functioning, memory, complex cognition, social cognition, and sensorimotor speed. We examined relationships among specific and general psychopathology dimensions and neurocognitive factors.

RESULTS

The final model showed both overall and specific associations between cognitive functioning and psychopathology, with acceptable fit (CFI = 0.91; TLI = 0.90; RMSEA = 0.024; SRMR = 0.054). Overall psychopathology and most psychopathology dimensions were negatively associated with neurocognitive functioning (phobias [p < 0.0005], behavioral/externalizing [p < 0.0005], attention-deficit/hyperactivity [p < 0.0005], psychosis [p < 0.0005 to p < 0.05]), except for dysphoria/distress and obsessive-compulsive symptoms, which were positively associated with complex cognition (p < 0.05 and p < 0.01, respectively).

CONCLUSION

By modeling a broad range of cognitive and psychopathology domains in a large, diverse sample of youth, we found aspects of neurocognitive functioning shared across clinical phenotypes, as well as domain-specific patterns. Findings support transdiagnostic examination of cognitive performance to parse variability in the link between neurocognitive functioning and clinical phenotypes.

Network enrichment significance testing in brain-phenotype association studies

Functional networks often guide our interpretation of spatial maps of brain-phenotype associations. However, methods for assessing enrichment of associations within networks of interest have varied in terms of both scientific rigor and underlying assumptions. While some approaches have relied on subjective interpretations, others have made unrealistic assumptions about spatial properties of imaging data, leading to inflated false positive rates. We seek to address this gap in existing methodology by borrowing insight from a method widely used in genetics research for testing enrichment of associations between a set of genes and a phenotype of interest. We propose network enrichment significance testing (NEST), a flexible framework for testing the specificity of brain-phenotype associations to functional networks or other sub-regions of the brain. We apply NEST to study enrichment of associations with structural and functional brain imaging data from a large-scale neurodevelopmental cohort study.

A Demographic-Conditioned Variational Autoencoder for fMRI Distribution Sampling and Removal of Confounds

Objective

fMRI and derived measures such as functional connectivity (FC) have been used to predict brain age, general fluid intelligence, psychiatric disease status, and preclinical neurodegenerative disease. However, it is not always clear that all demographic confounds, such as age, sex, and race, have been removed from fMRI data. Additionally, many fMRI datasets are restricted to authorized researchers, making dissemination of these valuable data sources challenging.

Methods

We create a variational autoencoder (VAE)-based model, DemoVAE, to decorrelate fMRI features from demographics and generate high-quality synthetic fMRI data based on user-supplied demographics. We train and validate our model using two large, widely used datasets, the Philadelphia Neurodevel-opmental Cohort (PNC) and Bipolar and Schizophrenia Network for Intermediate Phenotypes (BSNIP).

Results

We find that DemoVAE recapitulates group differences in fMRI data while capturing the full breadth of individual variations. Significantly, we also find that most clinical and computerized battery fields that are correlated with fMRI data are not correlated with DemoVAE latents. An exception are several fields related to schizophrenia medication and symptom severity.

Conclusion

Our model generates fMRI data that captures the full distribution of FC better than traditional VAE or GAN models. We also find that most prediction using fMRI data is dependent on correlation with, and prediction of, demographics.

Significance

Our DemoVAE model allows for generation of high quality synthetic data conditioned on subject demographics as well as the removal of the confounding effects of demographics. We identify that FC-based prediction tasks are highly influenced by demographic confounds.

A Demographic-Conditioned Variational Autoencoder for fMRI Distribution Sampling and Removal of Confounds

Objective

fMRI and derived measures such as functional connectivity (FC) have been used to predict brain age, general fluid intelligence, psychiatric disease status, and preclinical neurodegenerative disease. However, it is not always clear that all demographic confounds, such as age, sex, and race, have been removed from fMRI data. Additionally, many fMRI datasets are restricted to authorized researchers, making dissemination of these valuable data sources challenging.

Methods

We create a variational autoencoder (VAE)-based model, DemoVAE, to decorrelate fMRI features from demographics and generate high-quality synthetic fMRI data based on user-supplied demographics. We train and validate our model using two large, widely used datasets, the Philadelphia Neurodevelopmental Cohort (PNC) and Bipolar and Schizophrenia Network for Intermediate Phenotypes (BSNIP).

Results

We find that DemoVAE recapitulates group differences in fMRI data while capturing the full breadth of individual variations. Significantly, we also find that most clinical and computerized battery fields that are correlated with fMRI data are not correlated with DemoVAE latents. An exception are several fields related to schizophrenia medication and symptom severity.

Conclusion

Our model generates fMRI data that captures the full distribution of FC better than traditional VAE or GAN models. We also find that most prediction using fMRI data is dependent on correlation with, and prediction of, demographics.

Significance

Our DemoVAE model allows for generation of high quality synthetic data conditioned on subject demographics as well as the removal of the confounding effects of demographics. We identify that FC-based prediction tasks are highly influenced by demographic confounds.

pNet: A toolbox for personalized functional networks modeling

Personalized functional networks (FNs) derived from functional magnetic resonance imaging (fMRI) data are useful for characterizing individual variations in the brain functional topography associated with the brain development, aging, and disorders. To facilitate applications of the personalized FNs with enhanced reliability and reproducibility, we develop an open-source toolbox that is user-friendly, extendable, and includes rigorous quality control (QC), featuring multiple user interfaces (graphics, command line, and a step-by-step guideline) and job-scheduling for high performance computing (HPC) clusters. Particularly, the toolbox, named personalized functional network modeling (pNet), takes fMRI inputs in either volumetric or surface type, ensuring compatibility with multiple fMRI data formats, and computes personalized FNs using two distinct modeling methods: one method optimizes the functional coherence of FNs, while the other enhances their independence. Additionally, the toolbox provides HTML-based reports for QC and visualization of personalized FNs. The toolbox is developed in both MATLAB and Python platforms with a modular design to facilitate extension and modification by users familiar with either programming language. We have evaluated the toolbox on two fMRI datasets and demonstrated its effectiveness and user-friendliness with interactive and scripting examples. pNet is publicly available at https://github.com/MLDataAnalytics/pNet.

Functional connectivity development along the sensorimotor-association axis enhances the cortical hierarchy

Human cortical maturation has been posited to be organized along the sensorimotor-association axis, a hierarchical axis of brain organization that spans from unimodal sensorimotor cortices to transmodal association cortices. Here, we investigate the hypothesis that the development of functional connectivity during childhood through adolescence conforms to the cortical hierarchy defined by the sensorimotor-association axis. We tested this pre-registered hypothesis in four large-scale, independent datasets (total n = 3355; ages 5-23 years): the Philadelphia Neurodevelopmental Cohort (n = 1207), Nathan Kline Institute-Rockland Sample (n = 397), Human Connectome Project: Development (n = 625), and Healthy Brain Network (n = 1126). Across datasets, the development of functional connectivity systematically varied along the sensorimotor-association axis. Connectivity in sensorimotor regions increased, whereas connectivity in association cortices declined, refining and reinforcing the cortical hierarchy. These consistent and generalizable results establish that the sensorimotor-association axis of cortical organization encodes the dominant pattern of functional connectivity development.

Interpretable Cognitive Ability Prediction: A Comprehensive Gated Graph Transformer Framework for Analyzing Functional Brain Networks

Graph convolutional deep learning has emerged as a promising method to explore the functional organization of the human brain in neuroscience research. This paper presents a novel framework that utilizes the gated graph transformer (GGT) model to predict individuals' cognitive ability based on functional connectivity (FC) derived from fMRI. Our framework incorporates prior spatial knowledge and uses a random-walk diffusion strategy that captures the intricate structural and functional relationships between different brain regions. Specifically, our approach employs learnable structural and positional encodings (LSPE) in conjunction with a gating mechanism to efficiently disentangle the learning of positional encoding (PE) and graph embeddings. Additionally, we utilize the attention mechanism to derive multi-view node feature embeddings and dynamically distribute propagation weights between each node and its neighbors, which facilitates the identification of significant biomarkers from functional brain networks and thus enhances the interpretability of the findings. To evaluate our proposed model in cognitive ability prediction, we conduct experiments on two large-scale brain imaging datasets: the Philadelphia Neurodevelopmental Cohort (PNC) and the Human Connectome Project (HCP). The results show that our approach not only outperforms existing methods in prediction accuracy but also provides superior explainability, which can be used to identify important FCs underlying cognitive behaviors.

Age Prediction Using Resting-State Functional MRI

The increasing lifespan and large individual differences in cognitive capability highlight the importance of comprehending the aging process of the brain. Contrary to visible signs of bodily ageing, like greying of hair and loss of muscle mass, the internal changes that occur within our brains remain less apparent until they impair function. Brain age, distinct from chronological age, reflects our brain's health status and may deviate from our actual chronological age. Notably, brain age has been associated with mortality and depression. The brain is plastic and can compensate even for severe structural damage by rewiring. Functional characterization offers insights that structural cannot provide. Contrary to the multitude of studies relying on structural magnetic resonance imaging (MRI), we utilize resting-state functional MRI (rsfMRI). We also address the issue of inclusion of subjects with abnormal brain ageing through outlier removal. In this study, we employ the Least Absolute Shrinkage and Selection Operator (LASSO) to identify the 39 most predictive correlations derived from the rsfMRI data. The data is from a cohort of 176 healthy right-handed volunteers, aged 18-78 years (95/81 male/female, mean age 48, SD 17) collected at the Mind Research Imaging Center at the National Cheng Kung University. We establish a normal reference model by excluding 68 outliers, which achieves a leave-one-out mean absolute error of 2.48 years. By asking which additional features that are needed to predict the chronological age of the outliers with a smaller error, we identify correlations predictive of abnormal aging. These are associated with the Default Mode Network (DMN). Our normal reference model has the lowest prediction error among published models evaluated on adult subjects of almost all ages and is thus a candidate for screening for abnormal brain aging that has not yet manifested in cognitive decline. This study advances our ability to predict brain aging and provides insights into potential biomarkers for assessing brain age, suggesting that the role of DMN in brain aging should be studied further.

In-vivo whole-cortex marker of excitation-inhibition ratio indexes cortical maturation and cognitive ability in youth

A balanced excitation-inhibition ratio (E/I ratio) is critical for healthy brain function. Normative development of cortex-wide E/I ratio remains unknown. Here we non-invasively estimate a putative marker of whole-cortex E/I ratio by fitting a large-scale biophysically-plausible circuit model to resting-state functional MRI (fMRI) data. We first confirm that our model generates realistic brain dynamics in the Human Connectome Project. Next, we show that the estimated E/I ratio marker is sensitive to the GABA-agonist benzodiazepine alprazolam during fMRI. Alprazolam-induced E/I changes are spatially consistent with positron emission tomography measurement of benzodiazepine receptor density. We then investigate the relationship between the E/I ratio marker and neurodevelopment. We find that the E/I ratio marker declines heterogeneously across the cerebral cortex during youth, with the greatest reduction occurring in sensorimotor systems relative to association systems. Importantly, among children with the same chronological age, a lower E/I ratio marker (especially in association cortex) is linked to better cognitive performance. This result is replicated across North American (8.2 to 23.0 years old) and Asian (7.2 to 7.9 years old) cohorts, suggesting that a more mature E/I ratio indexes improved cognition during normative development. Overall, our findings open the door to studying how disrupted E/I trajectories may lead to cognitive dysfunction in psychopathology that emerges during youth.

Somatomotor-visual resting state functional connectivity increases after 2 years in the UK Biobank longitudinal cohort

Purpose

Functional magnetic resonance imaging (fMRI) and functional connectivity (FC) have been used to follow aging in both children and older adults. Robust changes have been observed in children, in which high connectivity among all brain regions changes to a more modular structure with maturation. We examine FC changes in older adults after 2 years of aging in the UK Biobank (UKB) longitudinal cohort.

Approach

We process fMRI connectivity data using the Power264 atlas and then test whether the average internetwork FC changes in the 2722-subject longitudinal cohort are statistically significant using a Bonferroni-corrected t -test. We also compare the ability of Power264 and UKB-provided, independent component analysis (ICA)-based FC to determine which of a longitudinal scan pair is older. Finally, we investigate cross-sectional FC changes as well as differences due to differing scanner tasks in the UKB, Philadelphia Neurodevelopmental Cohort, and Alzheimer's Disease Neuroimaging Initiative datasets.

Results

We find a 6.8% average increase in somatomotor network (SMT)-visual network (VIS) connectivity from younger to older scans (corrected p < 10 - 15 ) that occurs in male, female, older subject (> 65 years old), and younger subject (< 55 years old) groups. Among all internetwork connections, the average SMT-VIS connectivity is the best predictor of relative scan age. Using the full FC and a training set of 2000 subjects, one is able to predict which scan is older 82.5% of the time using either the full Power264 FC or the UKB-provided ICA-based FC.

Conclusions

We conclude that SMT-VIS connectivity increases with age in the UKB longitudinal cohort and that resting state FC increases with age in the UKB cross-sectional cohort.

Data leakage inflates prediction performance in connectome-based machine learning models

Predictive modeling is a central technique in neuroimaging to identify brain-behavior relationships and test their generalizability to unseen data. However, data leakage undermines the validity of predictive models by breaching the separation between training and test data. Leakage is always an incorrect practice but still pervasive in machine learning. Understanding its effects on neuroimaging predictive models can inform how leakage affects existing literature. Here, we investigate the effects of five forms of leakage-involving feature selection, covariate correction, and dependence between subjects-on functional and structural connectome-based machine learning models across four datasets and three phenotypes. Leakage via feature selection and repeated subjects drastically inflates prediction performance, whereas other forms of leakage have minor effects. Furthermore, small datasets exacerbate the effects of leakage. Overall, our results illustrate the variable effects of leakage and underscore the importance of avoiding data leakage to improve the validity and reproducibility of predictive modeling.

Regression and Alignment for Functional Data and Network Topology

In the brain, functional connections form a network whose topological organization can be described by graph-theoretic network diagnostics. These include characterizations of the community structure, such as modularity and participation coefficient, which have been shown to change over the course of childhood and adolescence. To investigate if such changes in the functional network are associated with changes in cognitive performance during development, network studies often rely on an arbitrary choice of pre-processing parameters, in particular the proportional threshold of network edges. Because the choice of parameter can impact the value of the network diagnostic, and therefore downstream conclusions, we propose to circumvent that choice by conceptualizing the network diagnostic as a function of the parameter. As opposed to a single value, a network diagnostic curve describes the connectome topology at multiple scales-from the sparsest group of the strongest edges to the entire edge set. To relate these curves to executive function and other covariates, we use scalar-on-function regression, which is more flexible than previous functional data-based models used in network neuroscience. We then consider how systematic differences between networks can manifest in misalignment of diagnostic curves, and consequently propose a supervised curve alignment method that incorporates auxiliary information from other variables. Our algorithm performs both functional regression and alignment via an iterative, penalized, and nonlinear likelihood optimization. The illustrated method has the potential to improve the interpretability and generalizability of neuroscience studies where the goal is to study heterogeneity among a mixture of function- and scalar-valued measures.

Brain-phenotype predictions can survive across diverse real-world data

Recent work suggests that machine learning models predicting psychiatric treatment outcomes based on clinical data may fail when applied to unharmonized samples. Neuroimaging predictive models offer the opportunity to incorporate neurobiological information, which may be more robust to dataset shifts. Yet, among the minority of neuroimaging studies that undertake any form of external validation, there is a notable lack of attention to generalization across dataset-specific idiosyncrasies. Research settings, by design, remove the between-site variations that real-world and, eventually, clinical applications demand. Here, we rigorously test the ability of a range of predictive models to generalize across three diverse, unharmonized samples: the Philadelphia Neurodevelopmental Cohort (n=1291), the Healthy Brain Network (n=1110), and the Human Connectome Project in Development (n=428). These datasets have high inter-dataset heterogeneity, encompassing substantial variations in age distribution, sex, racial and ethnic minority representation, recruitment geography, clinical symptom burdens, fMRI tasks, sequences, and behavioral measures. We demonstrate that reproducible and generalizable brain-behavior associations can be realized across diverse dataset features with sample sizes in the hundreds. Results indicate the potential of functional connectivity-based predictive models to be robust despite substantial inter-dataset variability. Notably, for the HCPD and HBN datasets, the best predictions were not from training and testing in the same dataset (i.e., cross-validation) but across datasets. This result suggests that training on diverse data may improve prediction in specific cases. Overall, this work provides a critical foundation for future work evaluating the generalizability of neuroimaging predictive models in real-world scenarios and clinical settings.

The effects of data leakage on connectome-based machine learning models

Predictive modeling has now become a central technique in neuroimaging to identify complex brain-behavior relationships and test their generalizability to unseen data. However, data leakage, which unintentionally breaches the separation between data used to train and test the model, undermines the validity of predictive models. Previous literature suggests that leakage is generally pervasive in machine learning, but few studies have empirically evaluated the effects of leakage in neuroimaging data. Although leakage is always an incorrect practice, understanding the effects of leakage on neuroimaging predictive models provides insight into the extent to which leakage may affect the literature. Here, we investigated the effects of leakage on machine learning models in two common neuroimaging modalities, functional and structural connectomes. Using over 400 different pipelines spanning four large datasets and three phenotypes, we evaluated five forms of leakage fitting into three broad categories: feature selection, covariate correction, and lack of independence between subjects. As expected, leakage via feature selection and repeated subjects drastically inflated prediction performance. Notably, other forms of leakage had only minor effects (e.g., leaky site correction) or even decreased prediction performance (e.g., leaky covariate regression). In some cases, leakage affected not only prediction performance, but also model coefficients, and thus neurobiological interpretations. Finally, we found that predictive models using small datasets were more sensitive to leakage. Overall, our results illustrate the variable effects of leakage on prediction pipelines and underscore the importance of avoiding data leakage to improve the validity and reproducibility of predictive modeling.

Development of white matter fiber covariance networks supports executive function in youth

During adolescence, the brain undergoes extensive changes in white matter structure that support cognition. Data-driven approaches applied to cortical surface properties have led the field to understand brain development as a spatially and temporally coordinated mechanism that follows hierarchically organized gradients of change. Although white matter development also appears asynchronous, previous studies have relied largely on anatomical tract-based atlases, precluding a direct assessment of how white matter structure is spatially and temporally coordinated. Harnessing advances in diffusion modeling and machine learning, we identified 14 data-driven patterns of covarying white matter structure in a large sample of youth. Fiber covariance networks aligned with known major tracts, while also capturing distinct patterns of spatial covariance across distributed white matter locations. Most networks showed age-related increases in fiber network properties, which were also related to developmental changes in executive function. This study delineates data-driven patterns of white matter development that support cognition.

Age-associated alterations in thalamocortical structural connectivity in youths with a psychosis-spectrum disorder

Psychotic symptoms typically emerge in adolescence. Age-associated thalamocortical connectivity differences in psychosis remain unclear. We analyzed diffusion-weighted imaging data from 1254 participants 8-23 years old (typically developing (TD):N = 626, psychosis-spectrum (PS): N = 329, other psychopathology (OP): N = 299) from the Philadelphia Neurodevelopmental Cohort. We modeled thalamocortical tracts using deterministic fiber tractography, extracted Q-Space Diffeomorphic Reconstruction (QSDR) and diffusion tensor imaging (DTI) measures, and then used generalized additive models to determine group and age-associated thalamocortical connectivity differences. Compared to other groups, PS exhibited thalamocortical reductions in QSDR global fractional anisotropy (GFA, p-values range = 3.0 × 10-6-0.05) and DTI fractional anisotropy (FA, p-values range = 4.2 × 10-4-0.03). Compared to TD, PS exhibited shallower thalamus-prefrontal age-associated increases in GFA and FA during mid-childhood, but steeper age-associated increases during adolescence. TD and OP exhibited decreases in thalamus-frontal mean and radial diffusivities during adolescence; PS did not. Altered developmental trajectories of thalamocortical connectivity may contribute to the disruptions observed in adults with psychosis.

Similarity-based multimodal regression

To better understand complex human phenotypes, large-scale studies have increasingly collected multiple data modalities across domains such as imaging, mobile health, and physical activity. The properties of each data type often differ substantially and require either separate analyses or extensive processing to obtain comparable features for a combined analysis. Multimodal data fusion enables certain analyses on matrix-valued and vector-valued data, but it generally cannot integrate modalities of different dimensions and data structures. For a single data modality, multivariate distance matrix regression provides a distance-based framework for regression accommodating a wide range of data types. However, no distance-based method exists to handle multiple complementary types of data. We propose a novel distance-based regression model, which we refer to as Similarity-based Multimodal Regression (SiMMR), that enables simultaneous regression of multiple modalities through their distance profiles. We demonstrate through simulation, imaging studies, and longitudinal mobile health analyses that our proposed method can detect associations between clinical variables and multimodal data of differing properties and dimensionalities, even with modest sample sizes. We perform experiments to evaluate several different test statistics and provide recommendations for applying our method across a broad range of scenarios.

ImageNomer: Description of a functional connectivity and omics analysis tool and case study identifying a race confound

Most packages for the analysis of fMRI-based functional connectivity (FC) and genomic data are used with a programming language interface, lacking an easy-to-navigate GUI frontend. This exacerbates two problems found in these types of data: demographic confounds and quality control in the face of high dimensionality of features. The reason is that it is too slow and cumbersome to use a programming interface to create all the necessary visualizations required to identify all correlations, confounding effects, or quality control problems in a dataset. FC in particular usually contains tens of thousands of features per subject, and can only be summarized and efficiently explored using visualizations. To remedy this situation, we have developed ImageNomer, a data visualization and analysis tool that allows inspection of both subject-level and cohort-level demographic, genomic, and imaging features. The software is Python-based, runs in a self-contained Docker image, and contains a browser-based GUI frontend. We demonstrate the usefulness of ImageNomer by identifying an unexpected race confound when predicting achievement scores in the Philadelphia Neurodevelopmental Cohort (PNC) dataset, which contains multitask fMRI and single nucleotide polymorphism (SNP) data of healthy adolescents. In the past, many studies have attempted to use FC to identify achievement-related features in fMRI. Using ImageNomer to visualize trends in achievement scores between races, we find a clear potential for confounding effects if race can be predicted using FC. Using correlation analysis in the ImageNomer software, we show that FCs correlated with Wide Range Achievement Test (WRAT) score are in fact more highly correlated with race. Investigating further, we find that whereas both FC and SNP (genomic) features can account for 10-15% of WRAT score variation, this predictive ability disappears when controlling for race. We also use ImageNomer to investigate race-FC correlation in the Bipolar and Schizophrenia Network for Intermediate Phenotypes (BSNIP) dataset. In this work, we demonstrate the advantage of our ImageNomer GUI tool in data exploration and confound detection. Additionally, this work identifies race as a strong confound in FC data and casts doubt on the possibility of finding unbiased achievement-related features in fMRI and SNP data of healthy adolescents.

Pathways from threat exposure to psychotic symptoms in youth: The role of emotion recognition bias and brain structure

BACKGROUND

Research supports an association between threatening experiences in childhood and psychosis. It is possible that early threat exposure disrupts the development of emotion recognition (specifically, producing a bias for facial expressions relating to threat) and the brain structures subserving it, contributing to psychosis development.

METHODS

Using data from the Philadelphia Neurodevelopmental Cohort, we examined associations between threat exposure and both the misattribution of facial expressions to fear/anger in an emotion recognition task, and gray matter volumes in key emotion processing regions. Our sample comprised youth with psychosis spectrum symptoms (N = 304), control youth (N = 787), and to evaluate specificity, youth with internalizing symptoms (N = 92). The moderating effects of group and sex were examined.

RESULTS

Both the psychosis spectrum and internalizing groups had higher levels of threat exposure than controls. In the total sample, threat exposure was associated with lower left medial prefrontal cortex (mPFC) volume but not misattributions to fear/anger. The effects of threat exposure did not significantly differ by group or sex.

CONCLUSIONS

The findings of this study provide evidence for an effect of threat exposure on mPFC morphology, but do not support an association between threat exposure and a recognition bias for threat-related expressions, that is particularly pronounced in psychosis. Future research should investigate factors linking transdiagnostic alterations related to threat exposure with psychotic symptoms, and attempt to clarify the mechanisms underpinning emotion recognition misattributions in threat-exposed youth.

Power and reproducibility in the external validation of brain-phenotype predictions

Identifying reproducible and generalizable brain-phenotype associations is a central goal of neuroimaging. Consistent with this goal, prediction frameworks evaluate brain-phenotype models in unseen data. Most prediction studies train and evaluate a model in the same dataset. However, external validation, or the evaluation of a model in an external dataset, provides a better assessment of robustness and generalizability. Despite the promise of external validation and calls for its usage, the statistical power of such studies has yet to be investigated. In this work, we ran over 60 million simulations across several datasets, phenotypes, and sample sizes to better understand how the sizes of the training and external datasets affect statistical power. We found that prior external validation studies used sample sizes prone to low power, which may lead to false negatives and effect size inflation. Furthermore, increases in the external sample size led to increased simulated power directly following theoretical power curves, whereas changes in the training dataset size offset the simulated power curves. Finally, we compared the performance of a model within a dataset to the external performance. The within-dataset performance was typically within r=0.2 of the cross-dataset performance, which could help decide how to power future external validation studies. Overall, our results illustrate the importance of considering the sample sizes of both the training and external datasets when performing external validation.

White matter functional gradients and their formation in adolescence

It is well known that functional magnetic resonance imaging (fMRI) is a widely used tool for studying brain activity. Recent research has shown that fluctuations in fMRI data can reflect functionally meaningful patterns of brain activity within the white matter. We leveraged resting-state fMRI from an adolescent population to characterize large-scale white matter functional gradients and their formation during adolescence. The white matter showed gray-matter-like unimodal-to-transmodal and sensorimotor-to-visual gradients with specific cognitive associations and a unique superficial-to-deep gradient with nonspecific cognitive associations. We propose two mechanisms for their formation in adolescence. One is a "function-molded" mechanism that may mediate the maturation of the transmodal white matter via the transmodal gray matter. The other is a "structure-root" mechanism that may support the mutual mediation roles of the unimodal and transmodal white matter maturation during adolescence. Thus, the spatial layout of the white matter functional gradients is in concert with the gray matter functional organization. The formation of the white matter functional gradients may be driven by brain anatomical wiring and functional needs.

A systematic review and meta-analysis of resting-state fMRI in anxiety disorders: Need for data sharing to move the field forward

Anxiety disorders are among the most prevalent psychiatric disorders. Neuroimaging findings remain uncertain, and resting state functional magnetic resonance (rs-fMRI) connectivity is of particular interest since it is a scalable functional imaging modality. Given heterogeneous past findings for rs-fMRI in anxious individuals, we characterize patterns across anxiety disorders by conducting a systematic review and meta-analysis. Studies were included if they contained at the time of scanning both a healthy group and a patient group. Due to insufficient study numbers, the quantitative meta-analysis only included seed-based studies. We performed an activation likelihood estimation (ALE) analysis that compared patients and healthy volunteers. All analyses were corrected for family-wise error with a cluster-level threshold of p < .05. Patients exhibited hypo-connectivity between the amygdala and the medial frontal gyrus, anterior cingulate cortex, and cingulate gyrus. This finding, however, was not robust to potential file-drawer effects. Though limited by strict inclusion criteria, our results highlight the heterogeneous nature of reported findings. This underscores the need for data sharing when attempting to detect reliable patterns of disruption in brain activity across anxiety disorders.

Transmit Radiant Individuality to Offspring (TRIO) study: investigating intergenerational transmission effects on brain development

Intergenerational transmission is a crucial aspect of human development. Although prior studies have demonstrated the continuity of psychopathology and maladaptive upbringing environments between parents and offspring, the underlying neurobiological mechanisms remain unclear. We have begun a novel neuroimaging research project, the Transmit Radiant Individuality to Offspring (TRIO) study, which focuses on biological parent-offspring trios. The participants of the TRIO study were Japanese parent-offspring trios consisting of offspring aged 10-40 and their biological mother and father. Structural and functional brain images of all participants were acquired using magnetic resonance imaging (MRI). Saliva samples were collected for DNA analysis. We obtained psychosocial information, such as intelligence, mental health problems, personality traits, and experiences during the developmental period from each parent and offspring in the same manner as much as possible. By April 2023, we completed data acquisition from 174 trios consisting of fathers, mothers, and offspring. The target sample size was 310 trios. However, we plan to conduct genetic and epigenetic analyses, and the sample size is expected to be expanded further while developing this project into a multi-site collaborative study in the future. The TRIO study can challenge the elucidation of the mechanism of intergenerational transmission effects on human development by collecting diverse information from parents and offspring at the molecular, neural, and behavioral levels. Our study provides interdisciplinary insights into how individuals' lives are involved in the construction of the lives of their descendants in the subsequent generation.

Dorsal visual stream and LIMK1: hemideletion, haplotype, and enduring effects in children with Williams syndrome

BACKGROUND

Williams syndrome (WS), a rare neurodevelopmental disorder caused by hemizygous deletion of ~ 25 genes from chromosomal band 7q11.23, affords an exceptional opportunity to study associations between a well-delineated genetic abnormality and a well-characterized neurobehavioral profile. Clinically, WS is typified by increased social drive (often termed "hypersociability") and severe visuospatial construction deficits. Previous studies have linked visuospatial problems in WS with alterations in the dorsal visual processing stream. We investigated the impacts of hemideletion and haplotype variation of LIMK1, a gene hemideleted in WS and linked to neuronal maturation and migration, on the structure and function of the dorsal stream, specifically the intraparietal sulcus (IPS), a region known to be altered in adults with WS.

METHODS

We tested for IPS structural and functional changes using longitudinal MRI in a developing cohort of children with WS (76 visits from 33 participants, compared to 280 visits from 94 typically developing age- and sex-matched participants) over the age range of 5-22. We also performed MRI studies of 12 individuals with rare, shorter hemideletions at 7q11.23, all of which included LIMK1. Finally, we tested for effects of LIMK1 variation on IPS structure and imputed LIMK1 expression in two independent cohorts of healthy individuals from the general population.

RESULTS

IPS structural (p < 10-4 FDR corrected) and functional (p < .05 FDR corrected) anomalies previously reported in adults were confirmed in children with WS, and, consistent with an enduring genetic mechanism, were stable from early childhood into adulthood. In the short hemideletion cohort, IPS deficits similar to those in WS were found, although effect sizes were smaller than those found in WS for both structural and functional findings. Finally, in each of the two general population cohorts stratified by LIMK1 haplotype, IPS gray matter volume (pdiscovery < 0.05 SVC, preplication = 0.0015) and imputed LIMK1 expression (pdiscovery = 10-15, preplication = 10-23) varied according to LIMK1 haplotype.

CONCLUSIONS

This work offers insight into neurobiological and genetic mechanisms responsible for the WS phenotype and also more generally provides a striking example of the mechanisms by which genetic variation, acting by means of molecular effects on a neural intermediary, can influence human cognition and, in some cases, lead to neurocognitive disorders.

Somatomotor-Visual Resting State Functional Connectivity Increases After Two Years in the UK Biobank Longitudinal Cohort

Purpose

Functional magnetic resonance imaging (fMRI) and functional connectivity (FC) have been used to follow aging in both children and older adults. Robust changes have been observed in children, where high connectivity among all brain regions changes to a more modular structure with maturation. In this work, we examine changes in FC in older adults after two years of aging in the UK Biobank longitudinal cohort.

Approach

We process data using the Power264 atlas, then test whether FC changes in the 2,722-subject longitudinal cohort are statistically significant using a Bonferroni-corrected t-test. We also compare the ability of Power264 and UKB-provided, ICA-based FC to determine which of a longitudinal scan pair is older.

Results

We find a 6.8% average increase in SMT-VIS connectivity from younger to older scan (from ρ = 0.39 to ρ = 0.42) that occurs in male, female, older subject (> 65 years old), and younger subject (< 55 years old) groups. Among all inter-network connections, this average SMT-VIS connectivity is the best predictor of relative scan age, accurately predicting which scan is older 57% of the time. Using the full FC and a training set of 2,000 subjects, one is able to predict which scan is older 82.5% of the time using either the full Power264 FC or the UKB-provided ICA-based FC.

Conclusions

We conclude that SMT-VIS connectivity increases in the longitudinal cohort, while resting state FC increases generally with age in the cross-sectional cohort. However, we consider the possibility of a change in resting state scanner task between UKB longitudinal data acquisitions.

Somatomotor-Visual Resting State Functional Connectivity Increases After Two Years in the UK Biobank Longitudinal Cohort

Purpose

Functional magnetic resonance imaging (fMRI) and functional connectivity (FC) have been used to follow aging in both children and older adults. Robust changes have been observed in children, where high connectivity among all brain regions changes to a more modular structure with maturation. In this work, we examine changes in FC in older adults after two years of aging in the UK Biobank longitudinal cohort.

Approach

We process data using the Power264 atlas, then test whether FC changes in the 2,722-subject longitudinal cohort are statistically significant using a Bonferroni-corrected t-test. We also compare the ability of Power264 and UKB-provided, ICA-based FC to determine which of a longitudinal scan pair is older.

Results

We find a 6.8% average increase in SMT-VIS connectivity from younger to older scan (from ρ = 0.39 to ρ = 0.42 ) that occurs in male, female, older subject (> 65 years old), and younger subject (< 55 years old) groups. Among all inter-network connections, this average SMT-VIS connectivity is the best predictor of relative scan age, accurately predicting which scan is older 57% of the time. Using the full FC and a training set of 2,000 subjects, one is able to predict which scan is older 82.5% of the time using either the full Power264 FC or the UKB-provided ICA-based FC.

Conclusions

We conclude that SMT-VIS connectivity increases in the longitudinal cohort, while resting state FC increases generally with age in the cross-sectional cohort. However, we consider the possibility of a change in resting state scanner task between UKB longitudinal data acquisitions.

Using network control theory to study the dynamics of the structural connectome

Network control theory (NCT) is a simple and powerful tool for studying how network topology informs and constrains dynamics. Compared to other structure-function coupling approaches, the strength of NCT lies in its capacity to predict the patterns of external control signals that may alter dynamics in a desired way. We have extensively developed and validated the application of NCT to the human structural connectome. Through these efforts, we have studied (i) how different aspects of connectome topology affect neural dynamics, (ii) whether NCT outputs cohere with empirical data on brain function and stimulation, and (iii) how NCT outputs vary across development and correlate with behavior and mental health symptoms. In this protocol, we introduce a framework for applying NCT to structural connectomes following two main pathways. Our primary pathway focuses on computing the control energy associated with transitioning between specific neural activity states. Our second pathway focuses on computing average controllability, which indexes nodes' general capacity to control dynamics. We also provide recommendations for comparing NCT outputs against null network models. Finally, we support this protocol with a Python-based software package called network control theory for python (nctpy).

Identifiability in Functional Connectivity May Unintentionally Inflate Prediction Results

Functional magnetic resonance (fMRI) is an invaluable tool in studying cognitive processes in vivo. Many recent studies use functional connectivity (FC), partial correlation connectivity (PC), or fMRI-derived brain networks to predict phenotypes with results that sometimes cannot be replicated. At the same time, FC can be used to identify the same subject from different scans with great accuracy. In this paper, we show a method by which one can unknowingly inflate classification results from 61% accuracy to 86% accuracy by treating longitudinal or contemporaneous scans of the same subject as independent data points. Using the UK Biobank dataset, we find one can achieve the same level of variance explained with 50 training subjects by exploiting identifiability as with 10,000 training subjects without double-dipping. We replicate this effect in four different datasets: the UK Biobank (UKB), the Philadelphia Neurodevelopmental Cohort (PNC), the Bipolar and Schizophrenia Network for Intermediate Phenotypes (BSNIP), and an OpenNeuro Fibromyalgia dataset (Fibro). The unintentional improvement ranges between 7% and 25% in the four datasets. Additionally, we find that by using dynamic functional connectivity (dFC), one can apply this method even when one is limited to a single scan per subject. One major problem is that features such as ROIs or connectivities that are reported alongside inflated results may confuse future work. This article hopes to shed light on how even minor pipeline anomalies may lead to unexpectedly superb results.

R-Mixup: Riemannian Mixup for Biological Networks

Biological networks are commonly used in biomedical and healthcare domains to effectively model the structure of complex biological systems with interactions linking biological entities. However, due to their characteristics of high dimensionality and low sample size, directly applying deep learning models on biological networks usually faces severe overfitting. In this work, we propose R-Mixup, a Mixup-based data augmentation technique that suits the symmetric positive definite (SPD) property of adjacency matrices from biological networks with optimized training efficiency. The interpolation process in R-Mixup leverages the log-Euclidean distance metrics from the Riemannian manifold, effectively addressing the swelling effect and arbitrarily incorrect label issues of vanilla Mixup. We demonstrate the effectiveness of R-Mixup with five real-world biological network datasets on both regression and classification tasks. Besides, we derive a commonly ignored necessary condition for identifying the SPD matrices of biological networks and empirically study its influence on the model performance. The code implementation can be found in Appendix E.

Explainable Multimodal Deep Dictionary Learning to Capture Developmental Differences From Three fMRI Paradigms

OBJECTIVE

Multimodal-based methods show great potential for neuroscience studies by integrating complementary information. There has been less multimodal work focussed on brain developmental changes.

METHODS

We propose an explainable multimodal deep dictionary learning method to uncover both the commonality and specificity of different modalities, which learns the shared dictionary and the modality-specific sparse representations based on the multimodal data and their encodings of a sparse deep autoencoder.

RESULTS

By regarding three fMRI paradigms collected during two tasks and resting state as modalities, we apply the proposed method on multimodal data to identify the brain developmental differences. The results show that the proposed model can not only achieve better performance in reconstruction, but also yield age-related differences in reoccurring patterns. Specifically, both children and young adults prefer to switch among states during two tasks while staying within a particular state during rest, but the difference is that children possess more diffuse functional connectivity patterns while young adults have more focused functional connectivity patterns.

CONCLUSION AND SIGNIFICANCE

To uncover the commonality and specificity of three fMRI paradigms to developmental differences, multimodal data and their encodings are used to train the shared dictionary and the modality-specific sparse representations. Identifying brain network differences helps to understand how the neural circuits and brain networks form and develop with age.

Individual differences in delay discounting are associated with dorsal prefrontal cortex connectivity in children, adolescents, and adults

Delay discounting is a measure of impulsive choice relevant in adolescence as it predicts many real-life outcomes, including obesity and academic achievement. However, resting-state functional networks underlying individual differences in delay discounting during youth remain incompletely described. Here we investigate the association between multivariate patterns of functional connectivity and individual differences in impulsive choice in a large sample of children, adolescents, and adults. A total of 293 participants (9-23 years) completed a delay discounting task and underwent 3T resting-state fMRI. A connectome-wide analysis using multivariate distance-based matrix regression was used to examine whole-brain relationships between delay discounting and functional connectivity. These analyses revealed that individual differences in delay discounting were associated with patterns of connectivity emanating from the left dorsal prefrontal cortex, a default mode network hub. Greater delay discounting was associated with greater functional connectivity between the dorsal prefrontal cortex and other default mode network regions, but reduced connectivity with regions in the dorsal and ventral attention networks. These results suggest delay discounting in children, adolescents, and adults is associated with individual differences in relationships both within the default mode network and between the default mode and networks involved in attentional and cognitive control.

Image harmonization: A review of statistical and deep learning methods for removing batch effects and evaluation metrics for effective harmonization

Magnetic resonance imaging and computed tomography from multiple batches (e.g. sites, scanners, datasets, etc.) are increasingly used alongside complex downstream analyses to obtain new insights into the human brain. However, significant confounding due to batch-related technical variation, called batch effects, is present in this data; direct application of downstream analyses to the data may lead to biased results. Image harmonization methods seek to remove these batch effects and enable increased generalizability and reproducibility of downstream results. In this review, we describe and categorize current approaches in statistical and deep learning harmonization methods. We also describe current evaluation metrics used to assess harmonization methods and provide a standardized framework to evaluate newly-proposed methods for effective harmonization and preservation of biological information. Finally, we provide recommendations to end-users to advocate for more effective use of current methods and to methodologists to direct future efforts and accelerate development of the field.

Depression pathophysiology, risk prediction of recurrence and comorbid psychiatric disorders using genome-wide analyses

Depression is a common psychiatric disorder and a leading cause of disability worldwide. Here we conducted a genome-wide association study meta-analysis of six datasets, including >1.3 million individuals (371,184 with depression) and identified 243 risk loci. Overall, 64 loci were new, including genes encoding glutamate and GABA receptors, which are targets for antidepressant drugs. Intersection with functional genomics data prioritized likely causal genes and revealed new enrichment of prenatal GABAergic neurons, astrocytes and oligodendrocyte lineages. We found depression to be highly polygenic, with ~11,700 variants explaining 90% of the single-nucleotide polymorphism heritability, estimating that >95% of risk variants for other psychiatric disorders (anxiety, schizophrenia, bipolar disorder and attention deficit hyperactivity disorder) were influencing depression risk when both concordant and discordant variants were considered, and nearly all depression risk variants influenced educational attainment. Additionally, depression genetic risk was associated with impaired complex cognition domains. We dissected the genetic and clinical heterogeneity, revealing distinct polygenic architectures across subgroups of depression and demonstrating significantly increased absolute risks for recurrence and psychiatric comorbidity among cases of depression with the highest polygenic burden, with considerable sex differences. The risks were up to 5- and 32-fold higher than cases with the lowest polygenic burden and the background population, respectively. These results deepen the understanding of the biology underlying depression, its disease progression and inform precision medicine approaches to treatment.

Dynamic weighted hypergraph convolutional network for brain functional connectome analysis

The hypergraph structure has been utilized to characterize the brain functional connectome (FC) by capturing the high order relationships among multiple brain regions of interest (ROIs) compared with a simple graph. Accordingly, hypergraph neural network (HGNN) models have emerged and provided efficient tools for hypergraph embedding learning. However, most existing HGNN models can only be applied to pre-constructed hypergraphs with a static structure during model training, which might not be a sufficient representation of the complex brain networks. In this study, we propose a dynamic weighted hypergraph convolutional network (dwHGCN) framework to consider a dynamic hypergraph with learnable hyperedge weights. Specifically, we generate hyperedges based on sparse representation and calculate the hyper similarity as node features. The hypergraph and node features are fed into a neural network model, where the hyperedge weights are updated adaptively during training. The dwHGCN facilitates the learning of brain FC features by assigning larger weights to hyperedges with higher discriminative power. The weighting strategy also improves the interpretability of the model by identifying the highly active interactions among ROIs shared by a common hyperedge. We validate the performance of the proposed model on two classification tasks with three paradigms functional magnetic resonance imaging (fMRI) data from Philadelphia Neurodevelopmental Cohort. Experimental results demonstrate the superiority of our proposed method over existing hypergraph neural networks. We believe our model can be applied to other applications in neuroimaging for its strength in representation learning and interpretation.

Development of Iron Status Measures during Youth: Associations with Sex, Neighborhood Socioeconomic Status, Cognitive Performance, and Brain Structure

BACKGROUND

Iron is essential to brain function, and iron deficiency during youth may adversely impact neurodevelopment. Understanding the developmental time course of iron status and its association with neurocognitive functioning is important for identifying windows for intervention.

OBJECTIVES

This study aimed to characterize developmental change in iron status and understand its association with cognitive performance and brain structure during adolescence using data from a large pediatric health network.

METHODS

This study included a cross-sectional sample of 4899 participants (2178 males; aged 8-22 y at the time of participation, M [SD] = 14.24 [3.7]) who were recruited from the Children's Hospital of Philadelphia network. Prospectively collected research data were enriched with electronic medical record data that included hematological measures related to iron status, including serum hemoglobin, ferritin, and transferrin (33,015 total samples). At the time of participation, cognitive performance was assessed using the Penn Computerized Neurocognitive Battery, and brain white matter integrity was assessed using diffusion-weighted MRI in a subset of individuals.

RESULTS

Developmental trajectories were characterized for all metrics and revealed that sex differences emerged after menarche such that females had reduced iron status relative to males [all R2partial > 0.008; all false discovery rates (FDRs) < 0.05]. Higher socioeconomic status was associated with higher hemoglobin concentrations throughout development (R2partial = 0.005; FDR < 0.001), and the association was greatest during adolescence. Higher hemoglobin concentrations were associated with better cognitive performance during adolescence (R2partial = 0.02; FDR < 0.001) and mediated the association between sex and cognition (mediation effect = -0.107; 95% CI: -0.191, -0.02). Higher hemoglobin concentration was also associated with greater brain white matter integrity in the neuroimaging subsample (R2partial = 0.06, FDR = 0.028).

CONCLUSIONS

Iron status evolves during youth and is lowest in females and individuals of low socioeconomic status during adolescence. Diminished iron status during adolescence has consequences for neurocognition, suggesting that this critical period of neurodevelopment may be an important window for intervention that has the potential to reduce health disparities in at-risk populations.

Hippocampal anterior- posterior shift in childhood and adolescence

Hippocampal-cortical networks play an important role in neurocognitive development. Applying the method of Connectivity-Based Parcellation (CBP) on hippocampal-cortical structural covariance (SC) networks computed from T1-weighted magnetic resonance images, we examined how the hippocampus differentiates into subregions during childhood and adolescence (N = 1105, 6-18 years). In late childhood, the hippocampus mainly differentiated along the anterior-posterior axis similar to previous reported functional differentiation patterns of the hippocampus. In contrast, in adolescence a differentiation along the medial-lateral axis was evident, reminiscent of the cytoarchitectonic division into cornu ammonis and subiculum. Further meta-analytical characterization of hippocampal subregions in terms of related structural co-maturation networks, behavioural and gene profiling suggested that the hippocampal head is related to higher order functions (e.g. language, theory of mind, autobiographical memory) in late childhood morphologically co-varying with almost the whole brain. In early adolescence but not in childhood, posterior subicular SC networks were associated with action-oriented and reward systems. The findings point to late childhood as an important developmental period for hippocampal head morphology and to early adolescence as a crucial period for hippocampal integration into action- and reward-oriented cognition. The latter may constitute a developmental feature that conveys increased propensity for addictive disorders.

Latent Similarity Identifies Important Functional Connections for Phenotype Prediction

OBJECTIVE

Endophenotypes such as brain age and fluid intelligence are important biomarkers of disease status. However, brain imaging studies to identify these biomarkers often encounter limited numbers of subjects but high dimensional imaging features, hindering reproducibility. Therefore, we develop an interpretable, multivariate classification/regression algorithm, called Latent Similarity (LatSim), suitable for small sample size but high feature dimension datasets.

METHODS

LatSim combines metric learning with a kernel similarity function and softmax aggregation to identify task-related similarities between subjects. Inter-subject similarity is utilized to improve performance on three prediction tasks using multi-paradigm fMRI data. A greedy selection algorithm, made possible by LatSim's computational efficiency, is developed as an interpretability method.

RESULTS

LatSim achieved significantly higher predictive accuracy at small sample sizes on the Philadelphia Neurodevelopmental Cohort (PNC) dataset. Connections identified by LatSim gave superior discriminative power compared to those identified by other methods. We identified 4 functional brain networks enriched in connections for predicting brain age, sex, and intelligence.

CONCLUSION

We find that most information for a predictive task comes from only a few (1-5) connections. Additionally, we find that the default mode network is over-represented in the top connections of all predictive tasks.

SIGNIFICANCE

We propose a novel prediction algorithm for small sample, high feature dimension datasets and use it to identify connections in task fMRI data. Our work can lead to new insights in both algorithm design and neuroscience research.

The effect of seed location on functional connectivity: evidence from an image-based meta-analysis

Introduction

Default mode network (DMN) is the most involved network in the study of brain development and brain diseases. Resting-state functional connectivity (rsFC) is the most used method to study DMN, but different studies are inconsistent in the selection of seed. To evaluate the effect of different seed selection on rsFC, we conducted an image-based meta-analysis (IBMA).

Methods

We identified 59 coordinates of seed regions of interest (ROIs) within the default mode network (DMN) from 11 studies (retrieved from Web of Science and Pubmed) to calculate the functional connectivity; then, the uncorrected t maps were obtained from the statistical analyses. The IBMA was performed with the t maps.

Results

We demonstrate that the overlap of meta-analytic maps across different seeds' ROIs within DMN is relatively low, which cautions us to be cautious with seeds' selection.

Discussion

Future studies using the seed-based functional connectivity method should take the reproducibility of different seeds into account. The choice of seed may significantly affect the connectivity results.

Activation of Internal Correctness Monitoring Circuitry in Youths With Psychosis Spectrum Symptoms

BACKGROUND

Self-directed performance monitoring is a critical contributor to cognitive performance and general functioning and is impacted by psychiatric symptoms and personality traits, but has been understudied in psychosis-risk states. We have shown that ventral striatum (VS) responds to correctness during cognitive tasks where no explicit feedback is required, and this intrinsic reinforcement response is reduced in schizophrenia.

METHODS

Here, we examined this phenomenon in youths (n = 796, age range 11-22 years) from the Philadelphia Neurodevelopmental Cohort (PNC) performing a working memory functional magnetic resonance imaging task. We hypothesized that VS would respond to internal correctness monitoring, while classic salience network regions, such as dorsal anterior cingulate cortex and anterior insular cortex, would reflect internal error monitoring and that these responses would increase with age. We expected that neurobehavioral measures of performance monitoring would be reduced in youths with subclinical psychosis spectrum features and would correlate with amotivation severity.

RESULTS

Supporting these hypotheses, we found correct>incorrect activation in VS and incorrect>correct activation in anterior cingulate cortex and anterior insular cortex. Furthermore, VS activation was positively correlated with age, reduced in youths with psychosis spectrum features, and inversely correlated with amotivation. However, these patterns were not significant in anterior cingulate cortex and anterior insular cortex.

CONCLUSIONS

These findings advance our understanding of the neural underpinnings of performance monitoring and its impairment in adolescents with psychosis spectrum features. Such understanding can facilitate investigation of the developmental trajectory of normative and aberrant performance monitoring; contribute to early identification of youths at elevated risk for poor academic, occupational, or psychiatric outcomes; and provide potential targets for therapeutic development.

ModelArray: An R package for statistical analysis of fixel-wise data

Diffusion MRI is the dominant non-invasive imaging method used to characterize white matter organization in health and disease. Increasingly, fiber-specific properties within a voxel are analyzed using fixels. While tools for conducting statistical analyses of fixel-wise data exist, currently available tools support only a limited number of statistical models. Here we introduce ModelArray, an R package for mass-univariate statistical analysis of fixel-wise data. At present, ModelArray supports linear models as well as generalized additive models (GAMs), which are particularly useful for studying nonlinear effects in lifespan data. In addition, ModelArray also aims for scalable analysis. With only several lines of code, even large fixel-wise datasets can be analyzed using a standard personal computer. Detailed memory profiling revealed that ModelArray required only limited memory even for large datasets. As an example, we applied ModelArray to fixel-wise data derived from diffusion images acquired as part of the Philadelphia Neurodevelopmental Cohort (n = 938). ModelArray revealed anticipated nonlinear developmental effects in white matter. Moving forward, ModelArray is supported by an open-source software development model that can incorporate additional statistical models and other imaging data types. Taken together, ModelArray provides a flexible and efficient platform for statistical analysis of fixel-wise data.