Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
Review
. 2018 Sep;3(9):742-753.
doi: 10.1016/j.bpsc.2018.03.015. Epub 2018 Apr 5.

Understanding the Emergence of Neuropsychiatric Disorders With Network Neuroscience

Affiliations
Free PMC article
Review

Understanding the Emergence of Neuropsychiatric Disorders With Network Neuroscience

Danielle S Bassett et al. Biol Psychiatry Cogn Neurosci Neuroimaging. .
Free PMC article

Abstract

Major neuropsychiatric disorders such as psychosis are increasingly acknowledged to be disorders of brain connectivity. Yet tools to map, model, predict, and change connectivity are difficult to develop, largely because of the complex, dynamic, and multivariate nature of interactions between brain regions. Network neuroscience (NN) provides a theoretical framework and mathematical toolset to address these difficulties. Building on areas of mathematics such as graph theory, NN in its simplest form summarizes neuroimaging data by treating brain regions as nodes in a graph and by treating interactions or connections between nodes as edges in the graph. Network metrics can then be used to quantitatively describe the architecture of the graph, which in turn reflects the network's function. We review evidence supporting the utility of NN in understanding psychiatric disorders, with a focus on normative brain network development and abnormalities associated with psychosis. We also emphasize relevant methodological challenges, such as motion artifact correction, which are particularly important to consider when applying network tools to developmental neuroimaging data. We close with a discussion of several emerging frontiers of NN in psychiatry, including generative network modeling and network control theory. We aim to offer an accessible introduction to this emerging field and motivate further work that uses NN to better understand the normative development of brain networks and alterations in that development that accompany or foreshadow psychiatric disease.

Keywords: Adolescence; Connectivity; Development; MRI; Network; Psychosis.

Conflict of interest statement

FINANCIAL DISCLOSURES

The authors report no biomedical financial interests or potential conflicts of interest.

Figures

Figure 1:
Figure 1:. A brief network primer.
a, The process of building a brain network begins with a parcellation of imaging voxels into regions of interest. Next, connections between those regions (estimated via functional connectivity, white matter tractography, or cross-subject covariance in morphometric variables) are summarized in an adjacency matrix, which in turn can be represented by a network or graph. b, A simple schematic of a few common graph metrics used to characterize human brain networks. First, node degree is given by the number of edges a node has. The clustering coefficient of a node in a binary graph can be defined as the number of triangles containing that node, divided by the number of connected triples containing that node. Motifs are subgraphs with a fixed pattern of connectivity: a triangle is an example of a motif, and a connected triple is another example of a motif. The shortest path length between two nodes is given by the smallest number of edges that must be traversed to get from one to the other. Core-periphery structure in a network is present when high degree nodes are also densely connected with one another (sometimes called a “rich club”), and when a periphery of low degree nodes preferentially connects to the core. Community structure in a network is present when the graph can be decomposed into modules, where nodes in one module are densely connected to one another but sparsely connected to nodes in other modules. Panel (b) is inspired by (137).
Figure 2.
Figure 2.. Modular segregation during adolescent development.
Consistent effects of neurodevelopment are apparent in both functional (panel a) and structural networks (panel b). Notably, connections that strengthen with age are much more likely to be within a network module, whereas connections that weaken with age predominantly span across modules. These edge level changes result in increased modular segregation of the network. Panel a adapted with permission from Satterthwaite et al., 2014; panels b and c adapted with permission from (61).
Figure 3.
Figure 3.. Psychosis is associated with alterations of network topology.
Convergent evidence across studies suggests abnormalities of normal developmental processes, including reduced connectivity within the network core and diminished segregation of network modules.
Figure 4.
Figure 4.. Effective de-noising limits associations between functional connectivity and inscanner motion.
a, When GSR is included in confound regression the distribution of correlations between motion and functional connectivity (“QC-FC”) are markedly reduced, and centered around zero. b, However, inclusion of GSR does result in a mild increase of distance dependence, which is quantified as the slope of the relationship between QC-FC correlations and the inter-node Euclidian distance. The confound regression model without GSR included 24 parameters, including 6 realignment parameters (3 rotations, 3 translations), as well as their temporal derivative, square, and square of their temporal derivatives. The confound regression model with GSR included 36 parameters, which included not just the 6 realignment parameters, but also the mean global signal, the mean white matter signal, and the mean CSF signal. These 9 base parameters were expanded as described above via their temporal derivative, square, and square of the temporal derivative. All data are drawn from a sample of 393 youth aged 8–22 and imaged as part of the Philadelphia Neurodevelopmental Cohort. All analyses include age and sex as covariates. Adapted with permission from (102).
Figure 5.
Figure 5.. Emerging Frontiers: Generative network modeling.
a, The space of generative models. Generative models can be differentiated from one another along several dimensions, one of which is the timescale over which they operate. A model’s timescale is related to its neurobiological plausibility. Models whose timescale is nearer to that of development can incorporate more realistic and interpretable features and, in turn, have the chance of uncovering realistic growth mechanisms. At the opposite end of the spectrum are “single shot” models, where connection probabilities are initialized early on and all connections and weights are generated in a single algorithmic step. Situated between these two extremes are growth models that exhibit intrinsic timescales over which connections and/or nodes are added to the network, but where the timescale has no clear biological interpretation. b, Summary of a geometric model for human white matter networks estimated from diffusion imaging. Observed (black) and synthetic (colors) networks generated at different points in a pre-defined parameter space of interest. Each of the model-generated synthetic networks was created using an edge addition algorithm, in which connections were added probabilistically and one at a time according to a set of parameterized wiring rules. c, Cumulative distributions of degree (orange), clustering coefficient (green), betweenness centrality (yellow), and edge length (purple) for the observed connectome (darker line) and best-fitting synthetic networks (lighter lines) for a representative participant. Panel (a) is adapted with permission from (119). Panels (b) and (c) are adapted with permission from (120).
Figure 6.
Figure 6.. Emerging Frontiers: Network control theory.
a, Network control theory provides statistics that can be calculated from a dynamical systems model of activity propagation along a fixed structural (or anatomical) network. Here we illustrate the notion of average controllability, which provides structural support for moving the brain to easy-to-reach states nearby on the theoretically constructed energy landscape, and the notion of modal controllability, which provides structural support for moving the brain to difficult-to-reach states, far away on the theoretically constructed energy landscape. b, Applying these notions to structural networks estimated from diffusion tractography applied to diffusion spectrum imaging data acquired in 882 youth between the ages of 8 yr and 22 yr, Tang et al. observed a heterogeneous spatial distribution of average controllability values across 234 cortical and subcortical regions defined by the Lausanne atlas (133). c, In the same data set, Tang et al. observed that average controllability increases appreciably with age, as did modal controllability, while synchronizability decreased with age (not shown). d, To determine whether these statistics were sufficient to explain the observed developmental arc of white matter maturation, Tang and colleagues performed a game theoretic rewiring procedure in which edges were rewired to advance the Pareto front (either increasing controllability and decreasing synchronizability or keeping these statistics constant). Letters beside the networks indicate their topology: ring lattice (R), regular lattice (L), modular (M), and small-world (S). e, The authors observed that the simulated evolutionary trajectories track the human brain data points well, suggesting that one mechanism of human brain development is the reconfiguration of white matter connectivity to increase the human’s ability to flexibly move between diverse brain states. Panels (a-e) adapted with permission from (133).

Similar articles

See all similar articles

Cited by 6 articles

See all "Cited by" articles

Publication types

MeSH terms

Feedback