Multivariate brain network graph identification in functional MRI

The multiverse of data preprocessing and analysis in graph-based fMRI: A systematic literature review of analytical choices fed into a decision support tool for informed analysis

The large number of different analytical choices used by researchers is partly responsible for the challenge of replication in neuroimaging studies. For an exhaustive robustness analysis, knowledge of the full space of analytical options is essential. We conducted a systematic literature review to identify the analytical decisions in functional neuroimaging data preprocessing and analysis in the emerging field of cognitive network neuroscience. We found 61 different steps, with 17 of them having debatable parameter choices. Scrubbing, global signal regression, and spatial smoothing are among the controversial steps. There is no standardized order in which different steps are applied, and the parameter settings within several steps vary widely across studies. By aggregating the pipelines across studies, we propose three taxonomic levels to categorize analytical choices: 1) inclusion or exclusion of specific steps, 2) parameter tuning within steps, and 3) distinct sequencing of steps. We have developed a decision support application with high educational value called METEOR to facilitate access to the data in order to design well-informed robustness (multiverse) analysis.

Beta-informativeness-diffusion multilayer graph embedding for brain network analysis

Brain network analysis provides essential insights into the diagnosis of brain disease. Integrating multiple neuroimaging modalities has been demonstrated to be more effective than using a single modality for brain network analysis. However, a majority of existing brain network analysis methods based on multiple modalities often overlook both complementary information and unique characteristics from various modalities. To tackle this issue, we propose the Beta-Informativeness-Diffusion Multilayer Graph Embedding (BID-MGE) method. The proposed method seamlessly integrates structural connectivity (SC) and functional connectivity (FC) to learn more comprehensive information for diagnosing neuropsychiatric disorders. Specifically, a novel beta distribution mapping function (beta mapping) is utilized to increase vital information and weaken insignificant connections. The refined information helps the diffusion process concentrate on crucial brain regions to capture more discriminative features. To maximize the preservation of the unique characteristics of each modality, we design an optimal scale multilayer brain network, the inter-layer connections of which depend on node informativeness. Then, a multilayer informativeness diffusion is proposed to capture complementary information and unique characteristics from various modalities and generate node representations by incorporating the features of each node with those of their connected nodes. Finally, the node representations are reconfigured using principal component analysis (PCA), and cosine distances are calculated with reference to multiple templates for statistical analysis and classification. We implement the proposed method for brain network analysis of neuropsychiatric disorders. The results indicate that our method effectively identifies crucial brain regions associated with diseases, providing valuable insights into the pathology of the disease, and surpasses other advanced methods in classification performance.

Estimation of Discriminative Multimodal Brain Network Connectivity Using Message-Passing-Based Nonlinear Network Fusion

Effective estimation of brain network connectivity enables better unraveling of the extraordinary complexity interactions of brain regions and helps in auxiliary diagnosis of psychiatric disorders. Considering different modalities can provide comprehensive characterizations of brain connectivity, we propose the message-passing-based nonlinear network fusion (MP-NNF) algorithm to estimate multimodal brain network connectivity. In the proposed method, the initial functional and structural networks were computed from fMRI and DTI separately. Then, we update every unimodal network iteratively, making it more similar to the others in every iteration, and finally converge to one unified network. The estimated brain connectivities integrate complementary information from multiple modalities while preserving their original structure, by adding the strong connectivities present in unimodal brain networks and eliminating the weak connectivities. The effectiveness of the method was evaluated by applying the learned brain connectivity for the classification of major depressive disorder (MDD). Specifically, 82.18% classification accuracy was achieved even with the simple feature selection and classification pipeline, which significantly outperforms the competing methods. Exploration of brain connectivity contributed to MDD identification suggests that the proposed method not only improves the classification performance but also was sensitive to critical disease-related neuroimaging biomarkers.

Robust Brain State Decoding using Bidirectional Long Short Term Memory Networks in functional MRI

Decoding brain states of the underlying cognitive processes via learning discriminative feature representations has recently gained a lot of interest in brain imaging studies. Particularly, there has been an impetus to encode the dynamics of brain functioning by analyzing temporal information available in the fMRI data. Long-short term memory (LSTM), a class of machine learning model possessing a "memory" component, to retain previously seen temporal information, is increasingly being observed to perform well in various applications with dynamic temporal behavior, including brain state decoding. Because of the dynamics and inherent latency in fMRI BOLD responses, future temporal context is crucial. However, it is neither encoded nor captured by the conventional LSTM model. This paper performs robust brain state decoding via information encapsulation from both the past and future instances of fMRI data via bi-directional LSTM. This allows for explicitly modeling the dynamics of BOLD response without any delay adjustment. To this end, we utilize a bidirectional LSTM, wherein, the input sequence is fed in normal time-order for one LSTM network, and in the reverse time-order, for another. The two hidden activations of forward and reverse directions in bi-LSTM are collated to build the "memory" of the model and are used to robustly predict the brain states at every time instance. Working memory data from the Human Connectome Project (HCP) is utilized for validation and was observed to perform 18% better than it's unidirectional counterpart in terms of accuracy in predicting the brain states.

Identification of Autism Spectrum Disorder With Functional Graph Discriminative Network

Autism spectrum disorder (ASD) is a specific brain disease that causes communication impairments and restricted interests. Functional connectivity analysis methodology is widely used in neuroscience research and shows much potential in discriminating ASD patients from healthy controls. However, due to heterogeneity of ASD patients, the performance of conventional functional connectivity classification methods is relatively poor. Graph neural network is an effective graph representation method to model structured data like functional connectivity. In this paper, we proposed a functional graph discriminative network (FGDN) for ASD classification. On the basis of pre-built graph templates, the proposed FGDN is able to effectively distinguish ASD patient from health controls. Moreover, we studied the size of training set for effective training, inter-site predictions, and discriminative brain regions. Discriminative brain regions were determined by the proposed model to investigate its applicability and biomarkers for ASD identification. For functional connectivity classification and analysis, FGDN is not only an effective tool for ASD identification but also a potential technique in neuroscience research.

Calcarine as a bridge between brain function and structure in irritable bowel syndrome: A multiplex network analysis

BACKGROUND AND AIM

Jointly analyzing structural and functional brain networks enables a better understanding of pathological underpinnings of irritable bowel syndrome (IBS). Multiplex network analysis provides a novel framework to study complex networks consisting of different types of connectivity patterns in multimodal data.

METHODS

In the present work, we integrated functional and structural networks to a multiplex network. Then, the multiplex metrics and the inner-layer/inter-layer hub nodes were investigated through 34 patients with IBS and 33 healthy controls.

RESULTS

Significantly differential multiplex degree in both left and right parts of calcarine was found, and meanwhile, IBS patients lost inner-layer hub properties in these regions. In addition, the left fusiform was no longer practicing as an inner-layer hub node, while the right median cingulate acted as a new inner-layer hub node in the IBS patients. Besides, the right calcarine, which lost its inner-layer hub identity, became a new inter-layer hub node, and the multiplex degree of the left hippocampus, which lost its inter-layer hub identity in IBS patients, was significantly positively correlated with the IBS Symptom Severity Score scores.

CONCLUSIONS

Inner-layer hub nodes of multiplex networks were preferentially vulnerable, and some inner-layer hub nodes would convert into inter-layer hub nodes in IBS patients. Besides, the inter-layer hub nodes might be influenced by IBS severity and therefore converted to general nodes.

Abnormal Functional Connectivity Between the Left Medial Superior Frontal Gyrus and Amygdala Underlying Abnormal Emotion and Premature Ejaculation: A Resting State fMRI Study

Introduction

Premature ejaculation (PE) is a common sexual dysfunction and is found to be associated with abnormal emotion. The amygdala plays an important role in the processing of emotion. The process of ejaculation is found to be mediated by the frontal-limbic neural circuits. However, the correlations between PE and emotion are still unclear.

Methods

Resting-state functional magnetic resonance imaging (rs-fMRI) data were acquired in 27 PE patients with stable emotion (SPE), 27 PE patients with abnormal emotion (NPE), and 30 healthy controls (HC). We used rs-fMRI to explore the underlying neural mechanisms in SPE, NPE, and HC by measuring the functional connectivity (FC). Differences of FC values among the three groups were compared when choosing bilateral amygdala as the regions of interest (ROIs). We also explored the correlations between the brain regions showing altered FC values and scores of the premature ejaculation diagnostic tool (PEDT)/Eysenck Personality Inventory about neuroticism (EPQ-N) in the PE group.

Results

When the left amygdala was chosen as the ROI, the SPE group exhibited an increased FC between the left medial superior frontal gyrus (SFGmed) and amygdala compared with the NPE or HC group. When the right amygdala was chosen as the ROI, the NPE group exhibited a decreased FC between the left SFGmed and right amygdala compared with the HC group. In addition, FC values of the left SFGmed had positive correlations with PEDT and negative correlations with EPQ-N scores in the PE group. Moreover, FC values of the left superior temporal gyrus had positive correlations with EPQ-N scores in the PE group.

Conclusion

The increased FC values between the left SFGmed and amygdala could reflect a compensatory cortical control mechanism with the effect of stabilized emotion in the limbic regions of PE patients. Abnormal FC between these brain regions could play a critical role in the physiopathology of PE and could help us in dividing PE into more subtypes.

Exploring MEG brain fingerprints: Evaluation, pitfalls, and interpretations

Individual characterization of subjects based on their functional connectome (FC), termed "FC fingerprinting", has become a highly sought-after goal in contemporary neuroscience research. Recent functional magnetic resonance imaging (fMRI) studies have demonstrated unique characterization and accurate identification of individuals as an accomplished task. However, FC fingerprinting in magnetoencephalography (MEG) data is still widely unexplored. Here, we study resting-state MEG data from the Human Connectome Project to assess the MEG FC fingerprinting and its relationship with several factors including amplitude- and phase-coupling functional connectivity measures, spatial leakage correction, frequency bands, and behavioral significance. To this end, we first employ two identification scoring methods, differential identifiability and success rate, to provide quantitative fingerprint scores for each FC measurement. Secondly, we explore the edgewise and nodal MEG fingerprinting patterns across the different frequency bands (delta, theta, alpha, beta, and gamma). Finally, we investigate the cross-modality fingerprinting patterns obtained from MEG and fMRI recordings from the same subjects. We assess the behavioral significance of FC across connectivity measures and imaging modalities using partial least square correlation analyses. Our results suggest that fingerprinting performance is heavily dependent on the functional connectivity measure, frequency band, identification scoring method, and spatial leakage correction. We report higher MEG fingerprinting performances in phase-coupling methods, central frequency bands (alpha and beta), and in the visual, frontoparietal, dorsal-attention, and default-mode networks. Furthermore, cross-modality comparisons reveal a certain degree of spatial concordance in fingerprinting patterns between the MEG and fMRI data, especially in the visual system. Finally, the multivariate correlation analyses show that MEG connectomes have strong behavioral significance, which however depends on the considered connectivity measure and temporal scale. This comprehensive, albeit preliminary investigation of MEG connectome test-retest identifiability offers a first characterization of MEG fingerprinting in relation to different methodological and electrophysiological factors and contributes to the understanding of fingerprinting cross-modal relationships. We hope that this first investigation will contribute to setting the grounds for MEG connectome identification.

Decreased dynamism of overlapping brain sub-networks in Major Depressive Disorder

Major Depressive Disorder (MDD) is increasingly recognized as a common brain disorder with aberrant brain networks. Alterations in dynamic functional brain networks have been widely reported in MDD. However, previous studies mainly focused on detecting non-overlapping sub-networks/communities, neglecting the possibility that one brain region may belong to multiple sub-networks/communities. In the present work, we utilized tensor decomposition method to detect overlapping communities and study the dynamism of overlapping sub-networks through 58 patients with MDD and 63 age- and sex-matched healthy controls (HC). The strength vectors of communities were calculated and two-sample t-test was performed to investigate the statistical significance of the differences in dynamism of MDD and HC groups. We found that communities detected in two groups were pairwise region-matching but overlapped brain regions were almost totally different. We considered two region-matching communities in the two groups as a sub-network. Compared to HCs, MDD patients showed significantly decreased dynamism in five sub-networks which could be functionally mapped to Visual Network (VN), Default Mode Network (DMN), Cognitive Control Network (CCN), Bilateral Limbic Network (BLN) and Auditory Network (AN). The results showed that MDD might only have a marginal effect on the holistic detection of communities and the changes of overlapped brain regions in MDD patients might be put down to the alteration of hubs. Further statistical analysis on nine sub-networks showed decreased dynamism of five sub-networks in MDD patients, which might help us achieve a better understanding of mechanism in MDD.

A novel node-level structure embedding and alignment representation of structural networks for brain disease analysis

Brain networks based on various neuroimaging technologies, such as diffusion tensor image (DTI) and functional magnetic resonance imaging (fMRI), have been widely applied to brain disease analysis. Currently, there are several node-level structural measures (e.g., local clustering coefficients and node degrees) for representing and analyzing brain networks since they usually can reflect the topological structure of brain regions. However, these measures typically describe specific types of structural information, ignoring important network properties (i.e., small structural changes) that could further improve the performance of brain network analysis. To overcome this problem, in this paper, we first define a novel node-level structure embedding and alignment (nSEA) representation to accurately characterize the node-level structural information of the brain network. Different from existing measures that characterize a specific type of structural properties with a single value, our proposed nSEA method can learn a vector representation for each node, thus contain richer structure information to capture small structural changes. Furthermore, we develop an nSEA representation based learning (nSEAL) framework for brain disease analysis. Specifically, we first perform structural embedding to calculate node vector representations for each brain network and then align vector representations of all brain networks into the common space for two group-level network analyses, including a statistical analysis and brain disease classifications. Experiment results on a real schizophrenia dataset demonstrate that our proposed method not only discover disease-related brain regions that could help to better understand the pathology of brain diseases, but also improve the classification performance of brain diseases, compared with state-of-the-art methods.

Low rank and sparsity constrained method for identifying overlapping functional brain networks

Analysis of functional magnetic resonance imaging (fMRI) data has revealed that brain regions can be grouped into functional brain networks (fBNs) or communities. A community in fMRI analysis signifies a group of brain regions coupled functionally with one another. In neuroimaging, functional connectivity (FC) measure can be utilized to quantify such functionally connected regions for disease diagnosis and hence, signifies the need of devising novel FC estimation methods. In this paper, we propose a novel method of learning FC by constraining its rank and the sum of non-zero coefficients. The underlying idea is that fBNs are sparse and can be embedded in a relatively lower dimension space. In addition, we propose to extract overlapping networks. In many instances, communities are characterized as combinations of disjoint brain regions, although recent studies indicate that brain regions may participate in more than one community. In this paper, large-scale overlapping fBNs are identified on resting state fMRI data by employing non-negative matrix factorization. Our findings support the existence of overlapping brain networks.

Double temporal sparsity based accelerated reconstruction of compressively sensed resting-state fMRI

A number of reconstruction methods have been proposed recently for accelerated functional Magnetic Resonance Imaging (fMRI) data collection. However, existing methods suffer with the challenge of greater artifacts at high acceleration factors. This paper addresses the issue of accelerating fMRI collection via undersampled k-space measurements combined with the proposed method based on l¹-l¹ norm constraints, wherein we impose first l¹-norm sparsity on the voxel time series (temporal data) in the transformed domain and the second l¹-norm sparsity on the successive difference of the same temporal data. Hence, we name the proposed method as Double Temporal Sparsity based Reconstruction (DTSR) method. The robustness of the proposed DTSR method has been thoroughly evaluated both at the subject level and at the group level on real fMRI data. Results are presented at various acceleration factors. Quantitative analysis in terms of Peak Signal-to-Noise Ratio (PSNR) and other metrics, and qualitative analysis in terms of reproducibility of brain Resting State Networks (RSNs) demonstrate that the proposed method is accurate and robust. In addition, the proposed DTSR method preserves brain networks that are important for studying fMRI data. Compared to the existing methods, the DTSR method shows promising potential with an improvement of 10-12 dB in PSNR with acceleration factors upto 3.5 on resting state fMRI data. Simulation results on real data demonstrate that DTSR method can be used to acquire accelerated fMRI with accurate detection of RSNs.