Repeatability Analysis of Global and Local Metrics of Brain Structural Networks

Strong intercorrelations among global graph-theoretic indices of structural connectivity in the human brain

Graph-theoretic metrics derived from neuroimaging data have been heralded as powerful tools for uncovering neural mechanisms of psychological traits, psychiatric disorders, and neurodegenerative diseases. In N = 8,185 human structural connectomes from UK Biobank, we examined the extent to which 11 commonly-used global graph-theoretic metrics index distinct versus overlapping information with respect to interindividual differences in brain organization. Using unthresholded, FA-weighted networks we found that all metrics other than Participation Coefficient were highly intercorrelated, both with each other (mean |r| = 0.788) and with a topologically-naïve summary index of brain structure (mean edge weight; mean |r| = 0.873). In a series of sensitivity analyses, we found that overlap between metrics is influenced by the sparseness of the network and the magnitude of variation in edge weights. Simulation analyses representing a range of population network structures indicated that individual differences in global graph metrics may be intrinsically difficult to separate from mean edge weight. In particular, Closeness, Characteristic Path Length, Global Efficiency, Clustering Coefficient, and Small Worldness were nearly perfectly collinear with one another (mean |r| = 0.939) and with mean edge weight (mean |r| = 0.952) across all observed and simulated conditions. Global graph-theoretic measures are valuable for their ability to distill a high-dimensional system of neural connections into summary indices of brain organization, but they may be of more limited utility when the goal is to index separable components of interindividual variation in specific properties of the human structural connectome.

Structural filtering of functional data offered discriminative features for autism spectrum disorder

This study attempted to answer the question, "Can filtering the functional data through the frequency bands of the structural graph provide data with valuable features which are not valuable in unfiltered data"?. The valuable features discriminate between autism spectrum disorder (ASD) and typically control (TC) groups. The resting-state fMRI data was passed through the structural graph's low, middle, and high-frequency band (LFB, MFB, and HFB) filters to answer the posed question. The structural graph was computed using the diffusion tensor imaging data. Then, the global metrics of functional graphs and metrics of functional triadic interactions were computed for filtered and unfiltered rfMRI data. Compared to TCs, ASDs had significantly higher clustering coefficients in the MFB, higher efficiencies and strengths in the MFB and HFB, and lower small-world propensity in the HFB. These results show over-connectivity, more global integration, and decreased local specialization in ASDs compared to TCs. Triadic analysis showed that the numbers of unbalanced triads were significantly lower for ASDs in the MFB. This finding may indicate the reason for restricted and repetitive behavior in ASDs. Also, in the MFB and HFB, the numbers of balanced triads and the energies of triadic interactions were significantly higher and lower for ASDs, respectively. These findings may reflect the disruption of the optimum balance between functional integration and specialization. There was no significant difference between ASDs and TCs when using the unfiltered data. All of these results demonstrated that significant differences between ASDs and TCs existed in the MFB and HFB of the structural graph when analyzing the global metrics of the functional graph and triadic interaction metrics. Also, these results demonstrated that frequency bands of the structural graph could offer significant findings which were not found in the unfiltered data. In conclusion, the results demonstrated the promising perspective of using structural graph frequency bands for attaining discriminative features and new knowledge, especially in the case of ASD.

Impact of thresholding on the consistency and sensitivity of diffusion MRI-based brain networks in patients with cerebral small vessel disease

INTRODUCTION

Thresholding of low-weight connections of diffusion MRI-based brain networks has been proposed to remove false-positive connections. It has been previously established that this yields more reproducible scan-rescan network architecture in healthy subjects. In patients with brain disease, network measures are applied to assess inter-individual variation and changes over time. Our aim was to investigate whether thresholding also achieves improved consistency in network architecture in patients, while maintaining sensitivity to disease effects for these applications.

METHODS

We applied fixed-density and absolute thresholding on brain networks in patients with cerebral small vessel disease (SVD, n = 86; ≈24 months follow-up), as a clinically relevant exemplar condition. In parallel, we applied the same methods in healthy young subjects (n = 44; scan-rescan interval ≈4 months) as a frame of reference. Consistency of network architecture was assessed with dice similarity of edges and intraclass correlation coefficient (ICC) of edge-weights and hub-scores. Sensitivity to disease effects in patients was assessed by evaluating interindividual variation, changes over time, and differences between those with high and low white matter hyperintensity burden, using correlation analyses and mixed ANOVA.

RESULTS

Compared to unthresholded networks, both thresholding methods generated more consistent architecture over time in patients (unthresholded: dice = .70; ICC: .70-.78; thresholded: dice = .77; ICC: .73-.83). However, absolute thresholding created fragmented nodes. Similar observations were made in the reference group. Regarding sensitivity to disease effects in patients, fixed-density thresholds that were optimal in terms of consistency (densities: .10-.30) preserved interindividual variation in global efficiency and node strength as well as the sensitivity to detect effects of time and group. Absolute thresholding produced larger fluctuations of interindividual variation.

CONCLUSIONS

Our results indicate that thresholding of low-weight connections, particularly when using fixed-density thresholding, results in more consistent network architecture in patients with longer rescan intervals, while preserving sensitivity to disease effects.

Diffusion Tensor MRI Structural Connectivity and PET Amyloid Burden in Preclinical Autosomal Dominant Alzheimer Disease: The DIAN Cohort

Background Pathologic evidence of Alzheimer disease (AD) is detectable years before onset of clinical symptoms. Imaging-based identification of structural changes of the brain in people at genetic risk for early-onset AD may provide insights into how genes influence the pathologic cascade that leads to dementia. Purpose To assess structural connectivity differences in cortical networks between cognitively normal autosomal dominant Alzheimer disease (ADAD) mutation carriers versus noncarriers and to determine the cross-sectional relationship of structural connectivity and cortical amyloid burden with estimated years to symptom onset (EYO) of dementia in carriers. Materials and Methods In this exploratory analysis of a prospective trial, all participants enrolled in the Dominantly Inherited Alzheimer Network between January 2009 and July 2014 who had normal cognition at baseline, T1-weighted MRI scans, and diffusion tensor imaging (DTI) were analyzed. Amyloid PET imaging using Pittsburgh compound B was also analyzed for mutation carriers. Areas of the cerebral cortex were parcellated into three cortical networks: the default mode network, frontoparietal control network, and ventral attention network. The structural connectivity of the three networks was calculated from DTI. General linear models were used to examine differences in structural connectivity between mutation carriers and noncarriers and the relationship between structural connectivity, amyloid burden, and EYO in mutation carriers. Correlation network analysis was performed to identify clusters of related clinical and imaging markers. Results There were 30 mutation carriers (mean age ± standard deviation, 34 years ± 10; 17 women) and 38 noncarriers (mean age, 37 years ± 10; 20 women). There was lower structural connectivity in the frontoparietal control network in mutation carriers compared with noncarriers (estimated effect of mutation-positive status, -0.0266; P = .04). Among mutation carriers, there was a correlation between EYO and white matter structural connectivity in the frontoparietal control network (estimated effect of EYO, -0.0015, P = .01). There was no significant relationship between cortical global amyloid burden and EYO among mutation carriers (P > .05). Conclusion White matter structural connectivity was lower in autosomal dominant Alzheimer disease mutation carriers compared with noncarriers and correlated with estimated years to symptom onset. Clinical trial registration no. NCT00869817 © RSNA, 2021 Online supplemental material is available for this article. See also the editorial by McEvoy in this issue.

Mapping Structural Connectivity Using Diffusion MRI: Challenges and Opportunities

Diffusion MRI-based tractography is the most commonly-used technique when inferring the structural brain connectome, i.e., the comprehensive map of the connections in the brain. The utility of graph theory-a powerful mathematical approach for modeling complex network systems-for analyzing tractography-based connectomes brings important opportunities to interrogate connectome data, providing novel insights into the connectivity patterns and topological characteristics of brain structural networks. When applying this framework, however, there are challenges, particularly regarding methodological and biological plausibility. This article describes the challenges surrounding quantitative tractography and potential solutions. In addition, challenges related to the calculation of global network metrics based on graph theory are discussed.Evidence Level: 5Technical Efficacy: Stage 1.

Graph Theoretic Analysis of Brain Connectomics in Multiple Sclerosis: Reliability and Relationship with Cognition

Research suggests that disruption of brain networks might explain cognitive deficits in multiple sclerosis (MS). The reliability and effectiveness of graph theoretic network metrics as measures of cognitive performance were tested in 37 people with MS and 23 controls. Specifically, relationships with cognitive performance (linear regression against the paced auditory serial addition test-3 seconds [PASAT-3], symbol digit modalities test [SDMT], and attention network test) and 1-month reliability (using the intraclass correlation coefficient [ICC]) of network metrics were measured using both resting-state functional and diffusion magnetic resonance imaging data. Cognitive impairment was directly related to measures of brain network segregation and inversely related to network integration (prediction of PASAT-3 by small worldness, modularity, characteristic path length, R² = 0.55; prediction of SDMT by small worldness, global efficiency, and characteristic path length, R² = 0.60). Reliability of the measures for 1 month in a subset of nine participants was mostly rated as good (ICC >0.6) for both controls and MS patients in both functional and diffusion data, but was highly dependent on the chosen parcellation and graph density, with the 0.2–0.5 density range being the most reliable. This suggests that disrupted network organization predicts cognitive impairment in MS and its measurement is reliable for a 1-month period. These new findings support the hypothesis of network disruption as a major determinant of cognitive deficits in MS and the future possibility of the application of derived metrics as surrogate outcomes in trials of therapies for cognitive impairment.

Effect of Fixed-Density Thresholding on Structural Brain Networks: A Demonstration in Cerebral Small Vessel Disease

A popular solution to control for edge density variability in structural brain network analysis is to threshold the networks to a fixed density across all subjects. However, it remains unclear how this type of thresholding affects the basic network architecture in terms of edge weights, hub location, and hub connectivity and, especially, how it affects the sensitivity to detect disease-related abnormalities. We investigated these two questions in a cohort of patients with cerebral small vessel disease and age-matched controls. Brain networks were reconstructed from diffusion magnetic resonance imaging data using deterministic fiber tractography. Networks were thresholded to a fixed density by removing edges with the lowest number of streamlines. We compared edge length (mm), fractional anisotropy (FA), proportion of hub connections, and hub location between the unthresholded and the thresholded networks of each subject. Moreover, we compared weighted graph measures of global and local connectivity obtained from the (un)thresholded networks between patients and controls. We performed these analyses over a range of densities (2-20%). Results indicate that fixed-density thresholding disproportionally removes edges composed of long streamlines, but is independent of FA. The edges removed were not preferentially connected to hub or nonhub nodes. Over half of the original hubs were reproducible when networks were thresholded to a density ≥10%. Furthermore, the between-group differences in graph measures observed in the unthresholded network remained present after thresholding, irrespective of the chosen density. We therefore conclude that moderate fixed-density thresholds can successfully be applied to control for the effects of density in structural brain network analysis.

Alterations in White Matter Network and Microstructural Integrity Differentiate Parkinson's Disease Patients and Healthy Subjects

Parkinson’s disease (PD) is a neurodegenerative disease, neuropathologically characterized by progressive loss of neurons in distinct brain areas. We hypothesize that quantifiable network alterations are caused by neurodegeneration. The primary motivation of this study was to assess the specific network alterations in PD patients that are distinct but appear in conjunction with physiological aging. 178 subjects (130 females) stratified into PD patients, young, middle-aged and elderly healthy controls (age- and sex-matched with PD patients), were analyzed using 3D-T1 magnetization-prepared rapid gradient-echo (MPRAGE) and diffusion weighted images acquired in 3T MRI scanner. Diffusion modeling and probabilistic tractography analysis were applied for generating voxel-based connectivity index maps from each seed voxel. The obtained connectivity matrices were analyzed using graph theoretical tools for characterization of involved network. By network-based statistic (NBS) the interregional connectivity differences between the groups were assessed. Measures evaluating local diffusion properties for anisotropy and diffusivity were computed for characterization of white matter microstructural integrity. The graph theoretical analysis showed a significant decrease in distance measures – eccentricity and characteristic path length – in PD patients in comparison to healthy subjects. Both measures as well were lower in PD patients when compared to young and middle-aged healthy controls. NBS analysis demonstrated lowered structural connectivity in PD patients in comparison to young and middle-aged healthy subject groups, mainly in frontal, cingulate, olfactory, insula, thalamus, and parietal regions. These specific network differences were distinct for PD and were not observed between the healthy subject groups. Microstructural analysis revealed diffusivity alterations within the white matter tracts in PD patients, predominantly in the body, splenium and tapetum of corpus callosum, corticospinal tract, and corona radiata, which were absent in normal aging. The identified alterations of network connectivity presumably caused by neurodegeneration indicate the disruption in global network integration in PD patients. The microstructural changes identified within the white matter could endorse network reconfiguration. This study provides a clear distinction between the network changes occurring during aging and PD. This will facilitate a better understanding of PD pathophysiology and the direct link between white matter changes and their role in the restructured network topology.

Test-Retest Reliability of Graph Theoretic Metrics in Adolescent Brains

Graph theory analysis of structural brain networks derived from diffusion tensor imaging (DTI) has become a popular analytical method in neuroscience, enabling advanced investigations of neurological and psychiatric disorders. The purpose of this study was to investigate (1) the effects of edge weighting schemes and (2) the effects of varying interscan periods on graph metrics within the adolescent brain. We compared a binary (B) network definition with three weighting schemes: fractional anisotropy (FA), streamline count, and streamline count with density and length correction (SDL). Two commonly used global and two local graph metrics were examined. The analysis was conducted with two groups of adolescent volunteers who received DTI scans either 12 weeks apart (16.62 ± 1.10 years) or within the same scanning session (30 min apart) (16.65 ± 1.14 years). The intraclass correlation coefficient was used to assess test-retest reliability and the coefficient of variation (CV) was used to assess precision. On average, each edge scheme produced reliable results at both time intervals. Weighted measures outperformed binary measures, with SDL weights producing the most reliable metrics. All edge schemes except FA displayed high CV values, leaving FA as the only edge scheme that consistently showed high precision while also producing reliable results. Overall findings suggest that FA weights are more suited for DTI connectome studies in adolescents.

Reproducibility of structural brain connectivity and network metrics using probabilistic diffusion tractography

The structural connectivity network constructed using probabilistic diffusion tractography can be characterized by the network metrics. In this study, short-term test-retest reproducibility of structural networks and network metrics were evaluated on 30 subjects in terms of within- and between-subject coefficient of variance (CV_ws, CV_bs), and intra class coefficient (ICC) using various connectivity thresholds. The short-term reproducibility under various connectivity thresholds were also investigated when subject groups have same or different sparsity. In summary, connectivity threshold of 0.01 can exclude around 80% of the edges with CV_ws = 73.2 ± 37.7%, CV_bs = 119.3 ± 44.0% and ICC = 0.62 ± 0.19. The rest 20% edges have CV_ws < 45%, CV_bs < 90%, ICC = 0.75 ± 0.12. The presence of 1% difference in the sparsity can cause additional within-subject variations on network metrics. In conclusion, applying connectivity thresholds on structural network to exclude spurious connections for the network analysis should be considered as necessities. Our findings suggest that a connectivity threshold over 0.01 can be applied without significant effect on the short-term when network metrics are evaluated at the same sparsity in subject group. When the sparsity is not the same, the procedure of integration over various connectivity thresholds can provide reliable estimation of network metrics.

Network-based approaches to examining stress in the adolescent brain

Exposure to stress, particularly in periods of rapid brain maturation such as adolescence, can profoundly influence developmental processes that undergird the organization of structural and functional brain networks and that may mediate the association between stressful experiences and maladaptive outcomes. While studies in translational developmental neuroscience often focus on how specific brain regions or targeted connections are altered by stress and psychiatric disease, the emerging field of network science may be especially valuable for elucidating the impact of stress on the intricate connectomics of the adolescent brain. Here we review recent studies that use graph theory and other network science approaches to understand normative adolescent brain development, effects of childhood maltreatment on the brain, and disorders characterized by pathological responses to stress in adolescents. Overall, these studies demonstrate that graph theory can be useful in identifying and quantifying developmental processes related to segregation, integration, and localized hub influence that are affected by stress exposure and that may lead to psychopathology. Finally, we discuss limitations in the current application of graph theory in this area and suggest what we believe are important directions for future work.

Whole-Brain High-Resolution Structural Connectome: Inter-Subject Validation and Application to the Anatomical Segmentation of the Striatum

The present study describes extraction of high-resolution structural connectome (HRSC) in 99 healthy subjects, acquired and made available by the Human Connectome Project. Single subject connectomes were then registered to the common surface space to allow assessment of inter-individual reproducibility of this novel technique using a leave-one-out approach. The anatomic relevance of the surface-based connectome was examined via a clustering algorithm, which identified anatomic subdivisions within the striatum. The connectivity of these striatal subdivisions were then mapped on the cortical and other subcortical surfaces. Findings demonstrate that HRSC analysis is robust across individuals and accurately models the actual underlying brain networks related to the striatum. This suggests that this method has the potential to model and characterize the healthy whole-brain structural network at high anatomic resolution.

Overcoming the effects of false positives and threshold bias in graph theoretical analyses of neuroimaging data

Graph theory (GT) is a powerful framework for quantifying topological features of neuroimaging-derived functional and structural networks. However, false positive (FP) connections arise frequently and influence the inferred topology of networks. Thresholding is often used to overcome this problem, but an appropriate threshold often relies on a priori assumptions, which will alter inferred network topologies. Four common network metrics (global efficiency, mean clustering coefficient, mean betweenness and smallworldness) were tested using a model tractography dataset. It was found that all four network metrics were significantly affected even by just one FP. Results also show that thresholding effectively dampens the impact of FPs, but at the expense of adding significant bias to network metrics. In a larger number (n=248) of tractography datasets, statistics were computed across random group permutations for a range of thresholds, revealing that statistics for network metrics varied significantly more than for non-network metrics (i.e., number of streamlines and number of edges). Varying degrees of network atrophy were introduced artificially to half the datasets, to test sensitivity to genuine group differences. For some network metrics, this atrophy was detected as significant (p<0.05, determined using permutation testing) only across a limited range of thresholds. We propose a multi-threshold permutation correction (MTPC) method, based on the cluster-enhanced permutation correction approach, to identify sustained significant effects across clusters of thresholds. This approach minimises requirements to determine a single threshold a priori. We demonstrate improved sensitivity of MTPC-corrected metrics to genuine group effects compared to an existing approach and demonstrate the use of MTPC on a previously published network analysis of tractography data derived from a clinical population. In conclusion, we show that there are large biases and instability induced by thresholding, making statistical comparisons of network metrics difficult. However, by testing for effects across multiple thresholds using MTPC, true group differences can be robustly identified.

Reproducibility of graph-theoretic brain network metrics: a systematic review

This systematic review aimed to assess the reproducibility of graph-theoretic brain network metrics. Primary research studies of test-retest reliability conducted on healthy human subjects were included that quantified test-retest reliability using either the intraclass correlation coefficient (ICC) or the coefficient of variance. The MEDLINE, Web of Knowledge, Google Scholar, and OpenGrey databases were searched up to February 2014. Risk of bias was assessed with 10 criteria weighted toward methodological quality. Twenty-three studies were included in the review (n=499 subjects) and evaluated for various characteristics, including sample size (5-45), retest interval (<1 h to >1 year), acquisition method, and test-retest reliability scores. For at least one metric, ICCs reached the fair range (ICC 0.40-0.59) in one study, the good range (ICC 0.60-0.74) in five studies, and the excellent range (ICC>0.74) in 16 studies. Heterogeneity of methods prevented further quantitative analysis. Reproducibility was good overall. For the metrics having three or more ICCs reported for both functional and structural networks, six of seven were higher in structural networks, indicating that structural networks may be more reliable over time. The authors were also able to highlight and discuss a number of methodological factors affecting reproducibility.

Validation of network communicability metrics for the analysis of brain structural networks

Computational network analysis provides new methods to analyze the brain's structural organization based on diffusion imaging tractography data. Networks are characterized by global and local metrics that have recently given promising insights into diagnosis and the further understanding of psychiatric and neurologic disorders. Most of these metrics are based on the idea that information in a network flows along the shortest paths. In contrast to this notion, communicability is a broader measure of connectivity which assumes that information could flow along all possible paths between two nodes. In our work, the features of network metrics related to communicability were explored for the first time in the healthy structural brain network. In addition, the sensitivity of such metrics was analysed using simulated lesions to specific nodes and network connections. Results showed advantages of communicability over conventional metrics in detecting densely connected nodes as well as subsets of nodes vulnerable to lesions. In addition, communicability centrality was shown to be widely affected by the lesions and the changes were negatively correlated with the distance from lesion site. In summary, our analysis suggests that communicability metrics that may provide an insight into the integrative properties of the structural brain network and that these metrics may be useful for the analysis of brain networks in the presence of lesions. Nevertheless, the interpretation of communicability is not straightforward; hence these metrics should be used as a supplement to the more standard connectivity network metrics.