Discriminative Analysis of Schizophrenia Patients Using Topological Properties of Structural and Functional Brain Networks: A Multimodal Magnetic Resonance Imaging Study

Interest in the application of machine learning (ML) techniques to multimodal magnetic resonance imaging (MRI) data for the diagnosis of schizophrenia (SZ) at the individual level is growing. However, a few studies have applied the features of structural and functional brain networks derived from multimodal MRI data to the discriminative analysis of SZ patients at different clinical stages. In this study, 205 normal controls (NCs), 61 first-episode drug-naive SZ (FESZ) patients, and 79 chronic SZ (CSZ) patients were recruited. We acquired their structural MRI, diffusion tensor imaging, and resting-state functional MRI data and constructed brain networks for each participant, including the gray matter network (GMN), white matter network (WMN), and functional brain network (FBN). We then calculated 3 nodal properties for each brain network, including degree centrality, nodal efficiency, and betweenness centrality. Two classifications (SZ vs. NC and FESZ vs. CSZ) were performed using five ML algorithms. We found that the SVM classifier with the input features of the combination of nodal properties of both the GMN and FBN achieved the best performance to discriminate SZ patients from NCs [accuracy, 81.2%; area under the receiver operating characteristic curve (AUC), 85.2%; p < 0.05]. Moreover, the SVM classifier with the input features of the combination of the nodal properties of both the GMN and WMN achieved the best performance to discriminate FESZ from CSZ patients (accuracy, 86.2%; AUC, 92.3%; p < 0.05). Furthermore, the brain areas in the subcortical/cerebellum network and the frontoparietal network showed significant importance in both classifications. Together, our findings provide new insights to understand the neuropathology of SZ and further highlight the potential advantages of multimodal network properties for identifying SZ patients at different clinical stages.

Resting-State Functional Connectivity in the Dorsal Attention Network Relates to Behavioral Performance in Spatial Attention Tasks and May Show Task-Related Adaptation

Between-subject variability in cognitive performance has been related to inter-individual differences in functional brain networks. Targeting the dorsal attention network (DAN) we questioned (i) whether resting-state functional connectivity (FC) within the DAN can predict individual performance in spatial attention tasks and (ii) whether there is short-term adaptation of DAN-FC in response to task engagement. Twenty-seven participants first underwent resting-state fMRI (PRE run), they subsequently performed different tasks of spatial attention [including visual search (VS)] and immediately afterwards received another rs-fMRI (POST run). Intra- and inter-hemispheric FC between core hubs of the DAN, bilateral intraparietal sulcus (IPS) and frontal eye field (FEF), was analyzed and compared between PRE and POST. Furthermore, we investigated rs-fMRI-behavior correlations between the DAN-FC in PRE/POST and task performance parameters. The absolute DAN-FC did not change from PRE to POST. However, different significant rs-fMRI-behavior correlations were revealed for intra-/inter-hemispheric connections in the PRE and POST run. The stronger the FC between left FEF and IPS before task engagement, the better was the learning effect (improvement of reaction times) in VS (r = 0.521, p = 0.024). And the faster the VS (mean RT), the stronger was the FC between right FEF and IPS after task engagement (r = −0.502, p = 0.032). To conclude, DAN-FC relates to the individual performance in spatial attention tasks supporting the view of functional brain networks as priors for cognitive ability. Despite a high inter- and intra-individual stability of DAN-FC, the change of FC-behavior correlations after task performance possibly indicates task-related adaptation of the DAN, underlining that behavioral experiences may shape intrinsic brain activity. However, spontaneous state fluctuations of the DAN-FC over time cannot be fully ruled out as an alternative explanation.

Embedding Functional Brain Networks in Low Dimensional Spaces Using Manifold Learning Techniques

Background: fMRI data is inherently high-dimensional and difficult to visualize. A recent trend has been to find spaces of lower dimensionality where functional brain networks can be projected onto manifolds as individual data points, leading to new ways to analyze and interpret the data. Here, we investigate the potential of two powerful non-linear manifold learning techniques for functional brain networks representation: (1) T-stochastic neighbor embedding (t-SNE) and (2) Uniform Manifold Approximation Projection (UMAP) a recent breakthrough in manifold learning.Methods: fMRI data from the Human Connectome Project (HCP) and an independent study of aging were used to generate functional brain networks. We used fMRI data collected during resting state data and during a working memory task. The relative performance of t-SNE and UMAP were investigated by projecting the networks from each study onto 2D manifolds. The levels of discrimination between different tasks and the preservation of the topology were evaluated using different metrics.Results: Both methods effectively discriminated the resting state from the memory task in the embedding space. UMAP discriminated with a higher classification accuracy. However, t-SNE appeared to better preserve the topology of the high-dimensional space. When networks from the HCP and aging studies were combined, the resting state and memory networks in general aligned correctly.Discussion: Our results suggest that UMAP, a more recent development in manifold learning, is an excellent tool to visualize functional brain networks. Despite dramatic differences in data collection and protocols, networks from different studies aligned correctly in the embedding space.

Atypical Resting State Functional Neural Network in Children With Autism Spectrum Disorder: Graph Theory Approach

Measuring whole brain networks is a promising approach to extract features of autism spectrum disorder (ASD), a brain disorder of widespread regions. Objectives of this study were to evaluate properties of resting-state functional brain networks in children with and without ASD and to evaluate their relation with social impairment severity. Magnetoencephalographic (MEG) data were recorded for 21 children with ASD (7 girls, 60–89 months old) and for 25 typically developing (TD) control children (10 girls, 60–91 months old) in a resting state while gazing at a fixation cross. After signal sources were localized onto the Desikan–Killiany brain atlas, statistical relations between localized activities were found and evaluated in terms of the phase lag index. After brain networks were constructed and after matching with intelligence using a coarsened exact matching algorithm, ASD and TD graph theoretical measures were compared. We measured autism symptoms severity using the Social Responsiveness Scale and investigated its relation with altered small-worldness using linear regression models. Children with ASD were found to have significantly lower small-worldness in the beta band (p = 0.007) than TD children had. Lower small-worldness in the beta band of children with ASD was associated with higher Social Responsiveness Scale total t-scores (p = 0.047). Significant relations were also inferred for the Social Awareness (p = 0.008) and Social Cognition (p = 0.015) sub-scales. Results obtained using graph theory demonstrate a difference between children with and without ASD in MEG-derived resting-state functional brain networks, and the relation of that difference with social impairment. Combining graph theory and MEG might be a promising approach to establish a biological marker for ASD.

Structure supports function: informing directed and dynamic functional connectivity with anatomical priors

The dynamic repertoire of functional brain networks is constrained by the underlying topology of structural connections. Despite this intrinsic relationship between structural (SC) and functional connectivity (FC), integrative and multimodal approaches to combine the two remain limited. Here, we propose a new adaptive filter for estimating dynamic and directed FC using structural connectivity information as priors. We tested the filter in rat epicranial recordings and human event-related EEG data, using SC priors from a meta-analysis of tracer studies and diffusion tensor imaging metrics, respectively. We show that, particularly under conditions of low signal-to-noise ratio, SC priors can help to refine estimates of directed FC, promoting sparse functional networks that combine information from structure and function. In addition, the proposed filter provides intrinsic protection against SC-related false negatives, as well as robustness against false positives, representing a valuable new tool for multimodal imaging in the context of dynamic and directed FC analysis.

Impaired dynamics of prefrontal beta-band networks during cognitive action control in Parkinson's disease

Among the cognitive symptoms that are associated with Parkinson's disease (PD), alterations in cognitive action control (CAC) are commonly reported in patients. CAC enables the suppression of an automatic action, in favor of a goal-directed one. The implementation of CAC is time-resolved and arguably associated with dynamic changes in functional brain networks. However, the electrophysiological functional networks involved, their dynamic changes, and how these changes are affected by PD, still remain unknown. In this study, to address this gap of knowledge, 21 PD patients and 10 healthy controls (HC) underwent a Simon task while high-density electroencephalography (HD-EEG) was recorded. Source-level dynamic connectivity matrices were estimated using the phase-locking value in the beta (12-25 Hz) and gamma (30-45 Hz) frequency bands. Temporal independent component analyses were used as a dimension reduction tool to isolate the group-specific brain network states that were dominant during the task. Typical microstate metrics were quantified to investigate the presence of these states at the subject-level. Our results first confirmed that PD patients experienced difficulties in inhibiting automatic responses during the task. At the group-level, HC displayed a significant functional network state that involved typical CAC-related prefrontal and cingulate nodes (e.g., inferior frontal cortex). Both group- and subject-level analyses showed that this network was less present in PD to the benefit of other networks involving lateralized temporal and insular components. The presence of this prefrontal network was associated with decreased reaction time. In the gamma band, two networks (fronto-cingulate and fronto-temporal) followed one another in HC, while 3 partially overlapping networks that included fronto-temporal, fronto-occipital and cross-hemispheric temporal connections were found in PD. At the subject-level, differences between PD and HC were less marked. Altogether, this study showed that the functional brain networks observed during CAC and their temporal changes were different in PD patients as compared to HC, and that these differences partially relate to behavioral changes. This study also highlights that task-based dynamic functional connectivity is a promising approach in understanding the cognitive dysfunctions observed in PD and beyond.

Associations of the Disrupted Functional Brain Network and Cognitive Function in End-Stage Renal Disease Patients on Maintenance Hemodialysis: A Graph Theory-Based Study of Resting-State Functional Magnetic Resonance Imaging

Objective: Cognitive impairment (CI) is a common neurological complication in patients with end-stage renal disease undergoing maintenance hemodialysis (MHD). Brain network analysis based on graph theory is a promising tool for studying CI. Therefore, the purpose of this study was to analyze the changes of functional brain networks in patients on MHD with and without CI by using graph theory and further explore the underlying neuropathological mechanism of CI in these patients.Methods: A total of 39 patients on MHD (19 cases with CI and 20 without) and 25 healthy controls (HCs) matched for age, sex, and years of education were enrolled in the study. Resting-state functional magnetic resonance imaging (rs-fMRI) and T1-weighted high-resolution anatomical data were obtained, and functional brain networks for each subject were constructed. The brain network parameters at the global and regional levels were calculated, and a one-way analysis of covariance was used to compare the differences across the three groups. The associations between the changed graph-theory parameters and cognitive function scores in patients on MHD were evaluated using Spearman correlation analysis.Results: Compared with HCs, the global parameters [sigma, gamma, and local efficiency (Eloc)] in both patient groups decreased significantly (p < 0.05, Bonferroni corrected). The clustering coefficient (Cp) in patients with CI was significantly lower than that in the other two groups (p < 0.05, Bonferroni corrected). The regional parameters were significantly lower in the right superior frontal gyrus, dorsolateral (SFGdor) and gyrus rectus (REC) of patients with CI than those of patients without CI; however the nodal local efficiency in the left amygdala was significantly increased (all p < 0.05, Bonferroni corrected). The global Cp and regional parameters in the three brain regions (right SFGdor, REC, and left amygdala) were significantly correlated with the cognitive function scores (all FDR q < 0.05).Conclusion: This study confirmed that the topology of the functional brain network was disrupted in patients on MHD with and without CI and the disruption of brain network was more severe in patients with CI. The abnormal brain network parameters are closely related to cognitive function in patients on MHD.

Vestibular Agnosia is linked to abnormal functional brain networks

Activation of the peripheral vestibular apparatus simultaneously elicits a reflex vestibular nystagmus and the vestibular perception of self-motion (vestibular-motion perception) or vertigo. In a newly characterised condition called Vestibular Agnosia found in conditions with disrupted brain network connectivity, e.g. traumatic brain injury (TBI) or neurodegeneration (Parkinson’s Disease), the link between vestibular reflex and perception is uncoupled, such that, peripheral vestibular activation elicits a vestibular ocular reflex nystagmus but without vertigo. Using structural brain imaging in acute traumatic brain injury, we recently linked vestibular agnosia to postural imbalance via disrupted right temporal white-matter circuits (inferior longitudinal fasciculus), however no white-matter tracts were specifically linked to vestibular agnosia. Given the relative difficulty in localizing the neuroanatomical correlates of vestibular-motion perception, and compatible with current theories of human consciousness (viz. the Global Neuronal Workspace Theory), we postulate that vestibular-motion perception (vertigo) is mediated by the coordinated interplay between fronto-parietal circuits linked to whole-brain broadcasting of the vestibular signal of self-motion. We thus used resting state functional MRI (rsfMRI) to map functional brain networks and hence test our postulate of an anterior-posterior cortical network mediating vestibular agnosia. Whole-brain rsfMRI was acquired from 39 prospectively recruited acute TBI patients (and 37 matched controls) with preserved peripheral and reflex vestibular function, along with self-motion perceptual thresholds during passive yaw rotations in the dark, and posturography. Following quality control of the brain imaging, 25 TBI patients’ images were analyzed. We classified 11 TBI patients with vestibular agnosia and 14 without vestibular agnosia based on laboratory testing of self-motion perception. Using independent component analysis, we found altered functional connectivity within posterior (right superior longitudinal fasciculus) and anterior networks (left rostral prefrontal cortex) in vestibular agnosia. Regions of interest analyses showed both inter-hemispheric and intra-hemispheric (left anterior-posterior) network disruption in vestibular agnosia. Assessing the brain regions linked via right inferior longitudinal fasciculus, a tract linked to vestibular agnosia in unbalanced patients (but now controlled for postural imbalance), seed-based analyses showed altered connectivity between higher order visual cortices involved in motion perception and mid-temporal regions. In conclusion, vestibular agnosia in our patient group is mediated by multiple brain network dysfunction, involving primarily left frontal and bilateral posterior networks. Understanding the brain mechanisms of vestibular agnosia provide both an insight into the physiological mechanisms of vestibular perception as well as an opportunity to diagnose and monitor vestibular cognitive deficits in brain disease such as TBI and neurodegeneration linked to imbalance and spatial disorientation.