The DIRECT consortium and the REST-meta-MDD project: towards neuroimaging biomarkers of major depressive disorder

Sex differences of signal complexity at resting-state functional magnetic resonance imaging and their associations with the estrogen-signaling pathway in the brain

Sex differences in the brain have been widely reported and may hold the key to elucidating sex differences in many medical conditions and drug response. However, the molecular correlates of these sex differences in structural and functional brain measures in the human brain remain unclear. Herein, we used sample entropy (SampEn) to quantify the signal complexity of resting-state functional magnetic resonance imaging (rsfMRI) in a large neuroimaging cohort (N = 1,642). The frontoparietal control network and the cingulo-opercular network had high signal complexity while the cerebellar and sensory motor networks had low signal complexity in both men and women. Compared with those in male brains, we found greater signal complexity in all functional brain networks in female brains with the default mode network exhibiting the largest sex difference. Using the gene expression data in brain tissues, we identified genes that were significantly associated with sex differences in brain signal complexity. The significant genes were enriched in the gene sets that were differentially expressed between the brain cortex and other tissues, the estrogen-signaling pathway, and the biological function of neural plasticity. In particular, the G-protein-coupled estrogen receptor 1 gene in the estrogen-signaling pathway was expressed more in brain regions with greater sex differences in SampEn. In conclusion, greater complexity in female brains may reflect the interactions between sex hormone fluctuations and neuromodulation of estrogen in women.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11571-023-09954-y.

The power of many brains: Catalyzing neuropsychiatric discovery through open neuroimaging data and large-scale collaboration

Recent advances in open neuroimaging data are enhancing our comprehension of neuropsychiatric disorders. By pooling images from various cohorts, statistical power has increased, enabling the detection of subtle abnormalities and robust associations, and fostering new research methods. Global collaborations in imaging have furthered our knowledge of the neurobiological foundations of brain disorders and aided in imaging-based prediction for more targeted treatment. Large-scale magnetic resonance imaging initiatives are driving innovation in analytics and supporting generalizable psychiatric studies. We also emphasize the significant role of big data in understanding neural mechanisms and in the early identification and precise treatment of neuropsychiatric disorders. However, challenges such as data harmonization across different sites, privacy protection, and effective data sharing must be addressed. With proper governance and open science practices, we conclude with a projection of how large-scale imaging resources and collaborations could revolutionize diagnosis, treatment selection, and outcome prediction, contributing to optimal brain health.

Spectral Graph Neural Network-Based Multi-Atlas Brain Network Fusion for Major Depressive Disorder Diagnosis

Major Depressive Disorder (MDD) imposes a substantial burden within the healthcare domain, impacting millions of individuals worldwide. Functional Magnetic Resonance Imaging (fMRI) has emerged as a promising tool for the objective diagnosis of MDD, enabling the investigation of functional connectivity patterns in the brain associated with this disorder. However, most existing methods focus on a single brain atlas, which limits their ability to capture the complex, multi-scale nature of functional brain networks. To address these limitations, we propose a novel multi-atlas fusion method that incorporates early and late fusion in a unified framework. Our method introduces the concept of the holistic Functional Connectivity Network (FCN), which captures both intra-atlas relationships within individual atlases and inter-regional relationships between atlases with different brain parcellation scales. This comprehensive representation enables the identification of potential disease-related patterns associated with MDD in the early stage of our framework. Moreover, by decoding the holistic FCN from various perspectives through multiple spectral Graph Convolutional Neural Networks and fusing their results with decision-level ensembles, we further improve the performance of MDD diagnosis. Our approach is easily implemented with minimal modifications to existing model structures and demonstrates a robust performance across different baseline models. Our method, evaluated on public resting-state fMRI datasets, surpasses the current multi-atlas fusion methods, enhancing the accuracy of MDD diagnosis. The proposed novel multi-atlas fusion framework provides a more reliable MDD diagnostic technique. Experimental results show our approach outperforms both single- and multi-atlas-based methods, demonstrating its effectiveness in advancing MDD diagnosis.

A Learnable Counter-Condition Analysis Framework for Functional Connectivity-Based Neurological Disorder Diagnosis

To understand the biological characteristics of neurological disorders with functional connectivity (FC), recent studies have widely utilized deep learning-based models to identify the disease and conducted post-hoc analyses via explainable models to discover disease-related biomarkers. Most existing frameworks consist of three stages, namely, feature selection, feature extraction for classification, and analysis, where each stage is implemented separately. However, if the results at each stage lack reliability, it can cause misdiagnosis and incorrect analysis in afterward stages. In this study, we propose a novel unified framework that systemically integrates diagnoses (i.e., feature selection and feature extraction) and explanations. Notably, we devised an adaptive attention network as a feature selection approach to identify individual-specific disease-related connections. We also propose a functional network relational encoder that summarizes the global topological properties of FC by learning the inter-network relations without pre-defined edges between functional networks. Last but not least, our framework provides a novel explanatory power for neuroscientific interpretation, also termed counter-condition analysis. We simulated the FC that reverses the diagnostic information (i.e., counter-condition FC): converting a normal brain to be abnormal and vice versa. We validated the effectiveness of our framework by using two large resting-state functional magnetic resonance imaging (fMRI) datasets, Autism Brain Imaging Data Exchange (ABIDE) and REST-meta-MDD, and demonstrated that our framework outperforms other competing methods for disease identification. Furthermore, we analyzed the disease-related neurological patterns based on counter-condition analysis.

Linking inter-subject variability of cerebellar functional connectome to clinical symptoms in major depressive disorder

Major depressive disorder (MDD) is a highly prevalent psychiatric disorder with remarkable inter-subject variability in clinical manifestations. Neuroimaging changes of the cerebellum have been recently proposed as a way to characterize MDD-related brain disruptions and might further explain various clinical symptoms. However, the cerebellar contributions to MDD clinical heterogeneity remain largely unknown. The analyzed data consisted of 251 MDD patients and 235 matching healthy controls (HC). The inter-subject variability of functional connectomes (IVFC) was estimated via Pearson's correlation analysis between each pair of the cerebellar and cerebral regions based on resting-state functional magnetic resonance imaging (rs-fMRI). A partial least squares (PLS) regression analysis was performed to determine the potential dimension linking the IVFC to clinical symptom measures. The results indicated that similar spatial distribution patterns of the cerebellar IVFC were observed between MDD and HC, but the MDD group exhibited abnormal IVFC alterations in the bilateral Cerebelum_4_5, bilateral Cerebelum_6, Vermis_1_2 and Vermis_8. The PLS model revealed that the IVFC pattern in the left Cerebelum_6 was significantly associated with three HAMD-17 items including the work and activities, psychomotor retardation, and depressed mood. These findings provided new evidence for the cerebellar changes in MDD. Specifically, we found that the altered inter-subject variability measurements correlated with clinical manifestations of this illness. Elucidating this variability could prove helpful for the evaluation of MDD heterogeneity as well as for understanding its pathophysiological mechanism.

Effective hyper-connectivity network construction and learning: Application to major depressive disorder identification

Functional connectivity (FC) derived from resting-state fMRI (rs-fMRI) is a primary approach for identifying brain diseases, but it is limited to capturing the pairwise correlation between regions-of-interest (ROIs) in the brain. Thus, hyper-connectivity which describes the higher-order relationship among multiple ROIs is receiving increasing attention. However, most hyper-connectivity methods overlook the directionality of connections. The direction of information flow constitutes a pivotal factor in shaping brain activity and cognitive processes. Neglecting this directional aspect can lead to an incomplete understanding of high-order interactions within the brain. To this end, we propose a novel effective hyper-connectivity (EHC) network that integrates direction detection and hyper-connectivity modeling. It characterizes the high-order directional information flow among multiple ROIs, providing a more comprehensive understanding of brain activity. Then, we develop a directed hypergraph convolutional network (DHGCN) to acquire deep representations from EHC network and functional indicators of ROIs. In contrast to conventional hypergraph convolutional networks designed for undirected hypergraphs, DHGCN is specifically tailored to handle directed hypergraph data structures. Moreover, unlike existing methods that primarily focus on fMRI time series, our proposed DHGCN model also incorporates multiple functional indicators, providing a robust framework for feature learning. Finally, deep representations generated via DHGCN, combined with demographic factors, are used for major depressive disorder (MDD) identification. Experimental results demonstrate that the proposed framework outperforms both FC and undirected hyper-connectivity models, as well as surpassing other state-of-the-art methods. The identification of EHC abnormalities through our framework can enhance the analysis of brain function in individuals with MDD.

Aberrant resting-state co-activation network dynamics in major depressive disorder

Major depressive disorder (MDD) is a globally prevalent and highly disabling disease characterized by dysfunction of large-scale brain networks. Previous studies have found that static functional connectivity is not sufficient to reflect the complicated and time-varying properties of the brain. The underlying dynamic interactions between brain functional networks of MDD remain largely unknown, and it is also unclear whether neuroimaging-based dynamic properties are sufficiently robust to discriminate individuals with MDD from healthy controls since the diagnosis of MDD mainly depends on symptom-based criteria evaluated by clinical observation. Resting-state functional magnetic resonance imaging (fMRI) data of 221 MDD patients and 215 healthy controls were shared by REST-meta-MDD consortium. We investigated the spatial-temporal dynamics of MDD using co-activation pattern analysis and made individual diagnoses using support vector machine (SVM). We found that MDD patients exhibited aberrant dynamic properties (such as dwell time, occurrence rate, transition probability, and entropy of Markov trajectories) in some transient networks including subcortical network (SCN), activated default mode network (DMN), de-activated SCN-cerebellum network, a joint network, activated attention network (ATN), and de-activated DMN-ATN, where some dynamic properties were indicative of depressive symptoms. The trajectories of other networks to deactivated DMN-ATN were more accessible in MDD patients. Subgroup analyses also showed subtle dynamic changes in first-episode drug-naïve (FEDN) MDD patients. Finally, SVM achieved preferable accuracies of 84.69%, 76.77%, and 88.10% in discriminating patients with MDD, FEDN MDD, and recurrent MDD from healthy controls with their dynamic metrics. Our findings reveal that MDD is characterized by aberrant dynamic fluctuations of brain network and the feasibility of discriminating MDD patients using dynamic properties, which provide novel insights into the neural mechanism of MDD.

Compressed primary-to-transmodal gradient is accompanied with subcortical alterations and linked to neurotransmitters and cellular signatures in major depressive disorder

Major depressive disorder (MDD) has been shown to involve widespread changes in low-level sensorimotor and higher-level cognitive functions. Recent research found that a primary-to-transmodal gradient could capture a cortical hierarchical organization ranging from perception and action to cognition in healthy subjects, but a prominent gradient dysfunction in MDD patients. However, whether and how this cortical gradient is linked to subcortical impairments and whether it is reflected in the microscale neurotransmitter systems and cell type-specific transcriptional signatures remain largely unknown. Data were acquired from 323 MDD patients and 328 sex- and age-matched healthy controls derived from the REST-meta-MDD project, and the human brain neurotransmitter systems density maps and gene expression data were drawn from two publicly available datasets. We investigated alterations of the primary-to-transmodal gradient in MDD patients and their correlations with clinical symptoms of depression and anxiety, as well as  their paralleled subcortical impairments. The correlations between MDD-related gradient alterations and densities of the neurotransmitter systems and gene expression information were assessed, respectively. The results demonstrated that MDD patients had a compressed primary-to-transmodal gradient accompanied by paralleled alterations in subcortical regions including the caudate, amygdala, and thalamus. The case-control gradient differences were spatially correlated with the densities of the neurotransmitter systems including the serotonin and dopamine receptors, and meanwhile with gene expression enriched in astrocytes, excitatory and inhibitory neuronal cells. These findings mapped the paralleled subcortical impairments in cortical hierarchical organization and also helped us understand the possible molecular and cellular substrates of the co-occurrence of high-level cognitive impairments with low-level sensorimotor abnormalities in MDD.

Dimensional subtyping of first-episode drug-naïve major depressive disorder: A multisite resting-state fMRI study

Major depressive disorder (MDD) is a heterogeneous syndrome, and understanding its neural mechanisms is crucial for the advancement of personalized medicine. However, conventional subtyping studies often categorize MDD patients into a single subgroup, neglecting the continuous interindividual variations. This implies a pressing need for a dimensional approach. 230 first-episode drug-naïve MDD patients and 395 healthy controls were obtained from 5 sites via the Rest-meta-MDD project. A Bayesian model was used to decompose the resting-state functional connectivity (RSFC) into multiple distinct RSFC patterns (refer to as "factors"), and each individual was allowed to express multiple factors to varying degrees (dimensional subtyping). The associations between demographic and clinical variables with the identified factors were calculated. We identified three latent factors with distinct but partially overlapping hypo- and hyper-RSFC patterns. Most participants co-expressed multiple latent factors. All factors shared abnormal RSFC involving the default mode network and frontoparietal network, but the directionality partially differed across factors. All factors were not significantly associated with demographic and clinical variables. These findings shed light on the interindividual variability in MDD and could form the basis for developing novel therapeutic approaches that capitalize on the heterogeneity of MDD.

Aberrant degree centrality profiles during rumination in major depressive disorder

Rumination is closely linked to the onset and maintenance of major depressive disorder (MDD). Prior neuroimaging studies have identified the association between self-reported rumination trait and the functional coupling among a network of brain regions using resting-state functional magnetic resonance imaging (MRI). However, little is known about the underlying neural circuitry mechanism during active rumination in MDD. Degree centrality (DC) is a simple metric to denote network integration, which is critical for higher-order psychological processes such as rumination. During an MRI scan, individuals with MDD (N = 45) and healthy controls (HC, N = 46) completed a rumination state task. We examined the interaction effect between the group (MDD vs. HC) and condition (rumination vs. distraction) on vertex-wise DC. We further characterized the identified brain region's functional involvement with Neurosynth and BrainMap. Network-wise seed-based functional connectivity (FC) analysis was also conducted for the identified region of interest. Finally, exploratory correlation analysis was conducted between the identified region of interest's network FCs and self-reported in-scanner affect levels. We found that a left superior frontal gyrus (SFG) region, generally overlapped with the frontal eye field, showed a significant interaction effect. Further analysis revealed its involvement with executive functions. FCs between this region, the frontoparietal, and the dorsal attention network (DAN) also showed significant interaction effects. Furthermore, its FC to DAN during distraction showed a marginally significant negative association with in-scanner affect level at the baseline. Our results implicated an essential role of the left SFG in the rumination's underlying neural circuitry mechanism in MDD and provided novel evidence for the conceptualization of rumination in terms of impaired executive control.

Normative modeling analysis reveals corpus callosum volume changes in early and mid-to-late first episode major depression

BACKGROUND

It has been widely accepted that major depressive disorder (MDD) impacts brain structures including the Corpus Callosum (CC). However, this assumption is based on scarce literature data involving small sample sizes. Furthermore, it is still unclear whether such CC volume changes may already be present at a first depressive episode.

METHODS

To further investigate this question, we compared 369 first-episode MDD patients (mean age = 35 years (sd = 12), 249 females; 283 early onset, 86 mid-to-late onset) from the open-source REST meta-MDD database closely matched for age and gender to 490 never-depressed individuals (mean age = 37 years (sd = 14); 309 females) using Z-scores obtained from normative neuroanatomical modeling to assess individual variability in CC (sub)volumes.

RESULTS

Relative to the norms established by the healthy controls, first-episode MDD patients displayed CC volume (z-score) reductions in the entire CC (including the body), as did mid-to-late-onset first-episode MDD patients (age ≥ 45 y). In early-onset first-episode MDD patients (age ≤ 44 y), depression severity symptoms were related to volume increases in the entire CC, as well as the body and splenium.

LIMITATIONS

No data on depressive episode duration. Relatively small sample size for mid-to-late first-episode MDD patients.

CONCLUSIONS

Our data revealed CC (sub)volume differences in early versus mid-to-late onset first episode MDD. Especially at early onset, depression severity may result in neural white matter activity as potential reaction to stress influences. Our results underline the importance of prompt clinical interventions at early onset MDD.

Altered topology of individual brain structural covariance networks in major depressive disorder

BACKGROUND

The neurobiological pathogenesis of major depression disorder (MDD) remains largely controversial. Previous literatures with limited sample size utilizing group-level structural covariance networks (SCN) commonly generated mixed findings regarding the topology of brain networks.

METHODS

We analyzed T1 images from a high-powered multisite sample including 1173 patients with MDD and 1019 healthy controls (HCs). We used regional gray matter volume to construct individual SCN by utilizing a novel approach based on the interregional effect size difference. We further investigated MDD-related structural connectivity alterations using topological metrics.

RESULTS

Compared to HCs, the MDD patients showed a shift toward randomization characterized by increased integration. Further subgroup analysis of patients in different stages revealed this randomization pattern was also observed in patients with recurrent MDD, while the first-episode drug naïve patients exhibited decreased segregation. Altered nodal properties in several brain regions which have a key role in both emotion regulation and executive control were also found in MDD patients compared with HCs. The abnormalities in inferior temporal gyrus were not influenced by any specific site. Moreover, antidepressants increased nodal efficiency in the anterior ventromedial prefrontal cortex.

CONCLUSIONS

The MDD patients at different stages exhibit distinct patterns of randomization in their brain networks, with increased integration during illness progression. These findings provide valuable insights into the disruption in structural brain networks that occurs in patients with MDD and might be useful to guide future therapeutic interventions.

Spatial-temporal data-augmentation-based functional brain network analysis for brain disorders identification

Introduction

Due to the lack of devices and the difficulty of gathering patients, the small sample size is one of the most challenging problems in functional brain network (FBN) analysis. Previous studies have attempted to solve this problem of sample limitation through data augmentation methods, such as sample transformation and noise addition. However, these methods ignore the unique spatial-temporal information of functional magnetic resonance imaging (fMRI) data, which is essential for FBN analysis.

Methods

To address this issue, we propose a spatial-temporal data-augmentation-based classification (STDAC) scheme that can fuse the spatial-temporal information, increase the samples, while improving the classification performance. Firstly, we propose a spatial augmentation module utilizing the spatial prior knowledge, which was ignored by previous augmentation methods. Secondly, we design a temporal augmentation module by random discontinuous sampling period, which can generate more samples than former approaches. Finally, a tensor fusion method is used to combine the features from the above two modules, which can make efficient use of spatial-temporal information of fMRI simultaneously. Besides, we apply our scheme to different types of classifiers to verify the generalization performance. To evaluate the effectiveness of our proposed scheme, we conduct extensive experiments on the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset and REST-meta-MDD Project (MDD) dataset.

Results

Experimental results show that the proposed scheme achieves superior classification accuracy (ADNI: 82.942%, MDD: 63.406%) and feature interpretation on the benchmark datasets.

Discussion

The proposed STDAC scheme, utilizing both spatial and temporal information, can generate more diverse samples than former augmentation methods for brain disorder classification and analysis.

Comprehensive evaluation of harmonization on functional brain imaging for multisite data-fusion

To embrace big-data neuroimaging, harmonizing the site effect in resting-state functional magnetic resonance imaging (R-fMRI) data fusion is a fundamental challenge. A comprehensive evaluation of potentially effective harmonization strategies, particularly with specifically collected data, has been scarce, especially for R-fMRI metrics. Here, we comprehensively assess harmonization strategies from multiple perspectives, including tests on residual site effect, individual identification, test-retest reliability, and replicability of group-level statistical results, on widely used R-fMRI metrics across various datasets, including data obtained from participants with repetitive measures at different scanners. For individual identifiability (i.e., whether the same subject could be identified across R-fMRI data scanned across different sites), we found that, while most methods decreased site effects, the Subsampling Maximum-mean-distance based distribution shift correction Algorithm (SMA) and parametric unadjusted CovBat outperformed linear regression models, linear mixed models, ComBat series and invariant conditional variational auto-encoder in clustering accuracy. Test-retest reliability was better for SMA and parametric adjusted CovBat than unadjusted ComBat series and parametric unadjusted CovBat in the number of overlapped voxels. At the same time, SMA was superior to the latter in replicability in terms of the Dice coefficient and the scale of brain areas showing sex differences reproducibly observed across datasets. Furthermore, SMA better detected reproducible sex differences of ALFF under the site-sex confounded situation. Moreover, we designed experiments to identify the best target site features to optimize SMA identifiability, test-retest reliability, and stability. We noted both sample size and distribution of the target site matter and introduced a heuristic formula for selecting the target site. In addition to providing practical guidelines, this work can inform continuing improvements and innovations in harmonizing methodologies for big R-fMRI data.

Gene Expression has Distinct Associations with Brain Structure and Function in Major Depressive Disorder

Major depressive disorder (MDD) is associated with structural and functional brain abnormalities. MDD as well as brain anatomy and function are influenced by genetic factors, but the role of gene expression remains unclear. Here, this work investigates how cortical gene expression contributes to structural and functional brain abnormalities in MDD. This work compares the gray matter volume and resting-state functional measures in a Chinese sample of 848 MDD patients and 749 healthy controls, and these case-control differences are then associated with cortical variation of gene expression. While whole gene expression is positively associated with structural abnormalities, it is negatively associated with functional abnormalities. This work observes the relationships of expression levels with brain abnormalities for individual genes, and found that transcriptional correlates of brain structure and function show opposite relations with gene dysregulation in postmortem cortical tissue from MDD patients. This work further identifies genes that are positively or negatively related to structural abnormalities as well as functional abnormalities. The MDD-related genes are enriched for brain tissue, cortical cells, and biological pathways. These findings suggest that distinct genetic mechanisms underlie structural and functional brain abnormalities in MDD, and highlight the importance of cortical gene expression for the development of cortical abnormalities.

The classification of brain network for major depressive disorder patients based on deep graph convolutional neural network

Introduction

The early diagnosis of major depressive disorder (MDD) is very important for patients that suffer from severe and irreversible consequences of depression. It has been indicated that functional connectivity (FC) analysis based on functional magnetic resonance imaging (fMRI) data can provide valuable biomarkers for clinical diagnosis. However, previous studies mainly focus on brain disease classification in small sample sizes, which may lead to dramatic divergences in classification accuracy.

Methods

This paper attempts to address this limitation by applying the deep graph convolutional neural network (DGCNN) method on a large multi-site MDD dataset. The resting-state fMRI data are acquired from 830 MDD patients and 771 normal controls (NC) shared by the REST-meta-MDD consortium.

Results

The DGCNN model trained with the binary network after thresholding, identified MDD patients from normal controls and achieved an accuracy of 72.1% with 10-fold cross-validation, which is 12.4%, 9.8%, and 7.6% higher than SVM, RF, and GCN, respectively. Moreover, the process of dataset reading and model training is faster. Therefore, it demonstrates the advantages of the DGCNN model with low time complexity and sound classification performance.

Discussion

Based on a large, multi-site dataset from MDD patients, the results expressed that DGCNN is not an extremely accurate method for MDD diagnosis. However, there is an improvement over previous methods with our goal of better understanding brain function and ultimately providing a biomarker or diagnostic capability for MDD diagnosis.