Fledgling pathoconnectomics of psychiatric disorders
2013; Elsevier BV; Volume: 17; Issue: 12 Linguagem: Inglês
10.1016/j.tics.2013.10.007
ISSN1879-307X
AutoresMikail Rubinov, Edward T. Bullmore,
Tópico(s)Functional Brain Connectivity Studies
Resumo•We evaluate the conceptual foundations of pathoconnectomics.•We overview the construction and analysis of empirical models of brain networks or connectomes.•We summarize recent reports of large-scale whole-brain connectome abnormalities of two candidate brain-network disorders, schizophrenia and autism. Pathoconnectomics, the mapping of abnormal brain networks, is a popular current framework for the study of brain dysfunction in psychiatric disorders. In this review we evaluate the conceptual foundations of this framework, describe the construction and analysis of empirical models of brain networks or connectomes, and summarize recent reports of the large-scale whole-brain connectome organization of two candidate brain-network disorders, schizophrenia and autism. We consider the evidence for the abnormal brain-network nature of psychiatric disorders and find it inconclusive. For instance, although there is some evidence for more random whole-brain network organization in schizophrenia and autism, future studies need to determine if these and other observed brain-network abnormalities represent sufficient phenotypes of psychiatric disorders, in order to validate pathoconnectomics as a scientific and clinical framework. Pathoconnectomics, the mapping of abnormal brain networks, is a popular current framework for the study of brain dysfunction in psychiatric disorders. In this review we evaluate the conceptual foundations of this framework, describe the construction and analysis of empirical models of brain networks or connectomes, and summarize recent reports of the large-scale whole-brain connectome organization of two candidate brain-network disorders, schizophrenia and autism. We consider the evidence for the abnormal brain-network nature of psychiatric disorders and find it inconclusive. For instance, although there is some evidence for more random whole-brain network organization in schizophrenia and autism, future studies need to determine if these and other observed brain-network abnormalities represent sufficient phenotypes of psychiatric disorders, in order to validate pathoconnectomics as a scientific and clinical framework. Connectomics, the mapping of brain networks (see Glossary), is a popular current framework for the study of brain function [1Sporns O. et al.The human connectome: a structural description of the human brain.PLoS Comput. Biol. 2005; 1: e42Crossref PubMed Scopus (2171) Google Scholar]. Connectomics postulates that brain functions, especially higher perceptual and cognitive functions, are contingent on brain-network interactions [2Sporns O. The human connectome: a complex network.Ann. N. Y. Acad. Sci. 2011; 1224: 109-125Crossref PubMed Scopus (942) Google Scholar, 3Alivisatos A.P. et al.The Brain Activity Map Project and the challenge of functional connectomics.Neuron. 2012; 74: 970-974Abstract Full Text Full Text PDF PubMed Scopus (379) Google Scholar] and that an understanding of these higher functions requires an understanding of brain-network organization [4Sporns O. Discovering the Human Connectome. MIT Press, 2012Crossref Google Scholar, 5Seung S. Connectome: How the Brain's Wiring Makes Us Who We Are. Houghton Mifflin Harcourt, 2012Google Scholar, 6Denk W. et al.Structural neurobiology: missing link to a mechanistic understanding of neural computation.Nat. Rev. Neurosci. 2012; 13: 351-358PubMed Google Scholar]. Abnormalities of higher brain functions are a prominent feature of major psychiatric disorders such as schizophrenia and autism. Pathoconnectomics, the mapping of abnormal brain networks, is a corollary framework of connectomics. Pathoconnectomics postulates that major psychiatric disorders are abnormalities of brain networks [7Insel T.R. Faulty circuits.Sci. Am. 2010; 302: 44-51Crossref PubMed Scopus (68) Google Scholar, 8Bullmore E. Sporns O. Complex brain networks: graph theoretical analysis of structural and functional systems.Nat. Rev. Neurosci. 2009; 10: 186-198Crossref PubMed Scopus (7508) Google Scholar] and that an understanding of these disorders requires an understanding of the corresponding abnormal brain-network organization [9Zorumski C.F. Rubin E.H. Psychiatry and Clinical Neuroscience: A Primer. Oxford University Press, 2011Crossref Google Scholar, 10Insel T. et al.Research domain criteria (RDoC): toward a new classification framework for research on mental disorders.Am. J. Psychiatry. 2010; 167: 748-751Crossref PubMed Scopus (4194) Google Scholar]. (We use the term pathoconnectomics for two reasons. First, this usage is consistent with past nomenclature, cf. 'pathophysiology of psychiatric disorders'. Second and more importantly, the mapping of brain dysfunction carries additional challenges to the mapping of healthy brain function and the usage of pathoconnectomics directly emphasizes this differentiation.) Pathoconnectomics is sometimes termed a new paradigm for the study of psychiatric disorders [11Kuhn T.S. The Structure of Scientific Revolutions. University of Chicago Press, 1996Crossref Google Scholar]. But the term paradigm has two distinct relevant meanings [12Kaiser D. In retrospect: the structure of scientific revolutions.Nature. 2012; 484: 164-166Google Scholar]. Pathoconnectomics is a paradigm in the sense of being a popular and disruptive framework [13Insel T.R. Disruptive insights in psychiatry: transforming a clinical discipline.J. Clin. Invest. 2009; 119: 700-705Crossref PubMed Scopus (115) Google Scholar]. But it is not a paradigm in the more important sense of being a significant scientific achievement; the framework is young and faces important challenges, some of which it shares with older branches of biological psychiatry. It remains to be seen whether pathoconnectomics provides anything close to approaching the explanatory power of other successful frameworks such as the neuron doctrine (the fundamental nature of the neuron as a unit of the nervous system [14Bullock T.H. et al.The neuron doctrine, redux.Science. 2005; 310: 791-793Crossref PubMed Scopus (124) Google Scholar]). The main challenges of pathoconnectomics are broadly twofold: a brain-network-based delineation of psychiatric disorders and an accurate definition of empirical models of brain networks. These challenges are notably interdependent: accurate empirical models of brain networks help to delineate psychiatric disorders and delineations of psychiatric disorders help to understand properties of brain networks important for higher brain function and dysfunction. Fulfillment of these challenges will allow a principled evaluation of the main tenet of pathoconnectomics, namely the abnormal brain-network nature of psychiatric disorders. But neglect of these challenges risks leading to a stagnant field of vague searches for unclear targets; similar problems affect other systems-biological investigations of complex disorders [15Buchanan A.V. et al.Dissecting complex disease: the quest for the Philosopher's Stone?.Int. J. Epidemiol. 2006; 35: 562-571Crossref PubMed Scopus (96) Google Scholar]. We now discuss these challenges in more detail. Objective delineation of psychiatric disorders is a central and perennial problem of psychiatry. In the current absence of such definitions, psychiatrists define psychiatric disorders using convenient, but not biologically validated, clinical phenotypes or groupings of symptoms and signs [16American Psychiatric Association Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5). American Psychiatric Publishing, 2013Google Scholar, 17World Health Organization International Statistical Classification of Diseases and Related Health Problems.10th edn. World Health Organization, 1992Google Scholar]. A biological phenotype objectively defines a disorder when it is specific for the disorder, such that its presence implies the presence of the disorder. Modern medicine uses the simplest-known specific biological phenotypes to define disorders [18Loscalzo J. Barabasi A-L. Systems biology and the future of medicine.Wiley Interdiscip. Rev. Syst. Biol. Med. 2011; 3: 619-627Crossref PubMed Scopus (196) Google Scholar]. Biological phenotypes that define disorders acquire primacy over clinical phenotypes of these disorders, such that clinical phenotypes are frequently altered to match biological phenotypes more closely. For instance, diabetes mellitus, a metabolic disorder, was initially defined by its clinical phenotype of voluminous urine output, weight loss, and thirst. The detection of elevated blood glucose as a specific phenotype helped to split diabetes mellitus from other disorders which have superficially similar clinical presentations, such as unrelated kidney diseases. Discoveries of more specific phenotypes continue to divide diabetes mellitus into further subgroups [19Tattersall R. Diabetes: The Biography. Oxford University Press, 2009Google Scholar]. This classification of disorders mirrors similar developments of scientific classification in other fields such as chemistry (of elements), biology (of organisms), and astronomy (of heavenly bodies) [20Bird A. Tobin E. Natural kinds.in: Zalta E.N. The Stanford Encyclopedia of Philosophy. Fall 2012. Standford University, 2012Google Scholar]. We use the term sufficient phenotype to denote the simplest-known specific biological phenotype of a disorder. We note that the main tenet of pathoconnectomics postulates that abnormal brain-networks are sufficient phenotypes of psychiatric disorders. We consider the available evidence for this tenet below. Psychiatric disorders associate with many genomic, proteomic, cellular, and systems phenotypes, including abnormalities of gray matter and white matter and functional activation and connectivity [21Hyman S.E. Can neuroscience be integrated into the DSM-V?.Nat. Rev. Neurosci. 2007; 8: 725-732Crossref PubMed Scopus (398) Google Scholar]. For instance, prominent early examples of abnormal brain structure and function include reduced gray-matter density of schizophrenia [22Shenton M.E. et al.A review of MRI findings in schizophrenia.Schizophr. Res. 2001; 49: 1-52Abstract Full Text Full Text PDF PubMed Scopus (1935) Google Scholar] and abnormal functional connectivity of autism [23Belmonte M.K. et al.Autism and abnormal development of brain connectivity.J. Neurosci. 2004; 24: 9228-9231Crossref PubMed Scopus (894) Google Scholar]. However, these associations are in most cases nonspecific. Psychiatric disorders also associate with abnormalities of brain networks, as we discuss below. But the presence of this association does not imply that psychiatric disorders should be viewed as abnormalities of brain networks, at least until such abnormalities are shown to represent sufficient phenotypes. This simple yet important fact is overlooked in the current discourse of pathoconnectomics. Biological psychiatry has made similar errors in the past, for instance by prematurely viewing schizophrenia and depression as disorders of dopamine and serotonin imbalances, respectively; these approaches have seemingly failed to yield major gains after several decades of research [24Kendler K.S. Schaffner K.F. The dopamine hypothesis of schizophrenia: an historical and philosophical analysis.Philos. Psychiatry Psychol. 2011; 18: 41-63Crossref Scopus (52) Google Scholar, 25Lacasse J.R. Leo J. Serotonin and depression: a disconnect between the advertisements and the scientific literature.PLoS Med. 2005; 2: e392Crossref PubMed Scopus (270) Google Scholar]. It would be useful for pathoconnectomics to avoid repeating these mistakes [26Editorial A critical look at connectomics.Nat. Neurosci. 2010; 13 (1441–1441)Google Scholar]. It is difficult to detect sufficient phenotypes of psychiatric disorders. One promising approach is to search for convergent effects of genes associated with these disorders. Major psychiatric disorders show moderate to high heritability and diverse genetic associations [27Cross-Disorder Group of the Psychiatric Genomics ConsortiumGenetic relationship between five psychiatric disorders estimated from genome-wide SNPs.Nat. Genet. 2013; 45: 984-994Crossref PubMed Scopus (1613) Google Scholar, 28Smoller J.W. et al.Identification of risk loci with shared effects on five major psychiatric disorders: a genome-wide analysis.Lancet. 2013; 381: 1371-1379Abstract Full Text Full Text PDF PubMed Scopus (2121) Google Scholar]. Genes associated with these disorders have heterogeneous functions in the nervous system; for instance, autism-associated genes modulate neuronal activity, cell adhesion, and activity-dependent protein synthesis [29Berg J.M. Geschwind D.H. Autism genetics: searching for specificity and convergence.Genome Biol. 2012; 13: 247Crossref PubMed Google Scholar]. The concept of an endophenotype is promising for identifying potential convergent effects of heterogeneous gene function. Endophenotypes are measurable and heritable (e.g., present at a higher rate in unaffected relatives) phenotypes of psychiatric disorders [30Miller G.A. Rockstroh B. Endophenotypes in psychopathology research: where do we stand?.Annu. Rev. Clin. Psychol. 2013; 9: 177-213Crossref PubMed Scopus (118) Google Scholar, 31Gottesman I.I. Gould T.D. The endophenotype concept in psychiatry: etymology and strategic intentions.Am. J. Psychiatry. 2003; 160: 636-645Crossref PubMed Scopus (4484) Google Scholar, 32Lenzenweger M.F. Endophenotype, intermediate phenotype, biomarker: definitions, concept comparisons, clarifications.Depress. Anxiety. 2013; 30: 185-189Crossref PubMed Scopus (66) Google Scholar]. Endophenotypes aim to identify genetically mediated traits that are simultaneously simpler than diverse genetic effects and more cohesive than heterogeneous clinical manifestations of disorders. There are similarities, but also important differences, between the concepts of sufficient phenotypes and endophenotypes. Most sufficient phenotypes are likely to be endophenotypes, but not all endophenotypes are sufficient phenotypes. In contrast to sufficient phenotypes, endophenotypes may include cognitive or behavioral traits and need not be simple or specific. Individual disorders may have many endophenotypes and an endophenotype may associate with many disorders. This lack of specificity makes endophenotypes easier to detect and usefully bypasses the subjective restrictions of psychiatric diagnostic classifications. The lack of specificity, however, also makes endophenotypes non-diagnostic. In the search for definitions of psychiatric disorders, endophenotypes serve as useful precursor traits to sufficient phenotypes. The connectome is broadly defined as the complete structural- or functional-network organization of the brain [1Sporns O. et al.The human connectome: a structural description of the human brain.PLoS Comput. Biol. 2005; 1: e42Crossref PubMed Scopus (2171) Google Scholar, 3Alivisatos A.P. et al.The Brain Activity Map Project and the challenge of functional connectomics.Neuron. 2012; 74: 970-974Abstract Full Text Full Text PDF PubMed Scopus (379) Google Scholar]. There are multiple microscopy- and neuroimaging-based model realizations of this concept (Table 1). Each of these empirical models has distinct spatial and sometimes temporal resolution, spatial coverage, and susceptibility to noise. The models balance the demands of biological realism and complexity. Neuronal-scale models may be too complex to construct and analyze, whereas regional-scale models may not be biologically realistic. Not all models are necessarily well suited for defining sufficient phenotypes of psychiatric disorders.Table 1Methods of structural connectomicsImaging methodApproximate spatial scaleEnvironment and example organismsAdvantagesDisadvantagesElectron microscopy∼NanometerEx vivo, roundworm, fruit flyAccurate characterization of dense neuronal and synaptic connectivitySmall volume of brain coverage, high computational costLight microscopy∼MicrometerEx vivo, mouseLarge volume coverage of neuronal projectionsInability to differentiate synapses and characterize dense connectivityMRI∼MillimeterIn vivo, monkey, humanWhole-brain coverage of large brainsInability to differentiate directionality, high susceptibility to noise Open table in a new tab Structural connectomes are maps of anatomical interactions between neural elements. Individual models differ on the spatial resolution and spatial extent of these maps. At the microscale, maps of synaptic connections between neurons represent the most intuitive representation of the structural connectome. High-resolution electron-microscopic and neuronal reconstruction techniques provide detailed neuronal and synaptic maps of these spatially dense neuronal circuits [33Helmstaedter M. Cellular-resolution connectomics: challenges of dense neural circuit reconstruction.Nat. Methods. 2013; 10: 501-507Crossref PubMed Scopus (197) Google Scholar]. These techniques were used to reconstruct the only currently-known complete synaptic wiring diagram of an organism, the roundworm Caenorhabditis elegans [34White J.G. et al.The structure of the nervous system of the nematode Caenorhabditis elegans.Philos. Trans. R. Soc. Lond. B: Biol. Sci. 1986; 314: 1-340Crossref PubMed Google Scholar]. The high computational cost and limited spatial extent of these techniques restrict their current use to mapping connectomes of small organisms. At the mesoscale, intermediate-resolution intermediate-extent light-microscopic and neural staining techniques provide whole-brain neuronal connection maps of larger organisms [35Osten P. Margrie T.W. Mapping brain circuitry with a light microscope.Nat. Methods. 2013; 10: 515-523Crossref PubMed Scopus (173) Google Scholar, 36Lichtman J.W. et al.A technicolour approach to the connectome.Nat. Rev. Neurosci. 2008; 9: 417-422Crossref PubMed Scopus (262) Google Scholar] and are currently used to derive connectome models of the mouse [37Bohland J.W. et al.A proposal for a coordinated effort for the determination of brainwide neuroanatomical connectivity in model organisms at a mesoscopic scale.PLoS Comput. Biol. 2009; 5: e1000334Crossref PubMed Scopus (216) Google Scholar]. The increase in volume of these models comes at the expense of sparser neuronal maps and reduced ability to infer the presence of neuronal connections. At the macroscale, low-resolution high-extent MRI techniques allow in vivo reconstruction of whole-brain connection maps of larger organisms, including humans [38Craddock R.C. et al.Imaging human connectomes at the macroscale.Nat. Methods. 2013; 10: 524-539Crossref PubMed Scopus (278) Google Scholar], and are presently used to construct a macroscale model of the connectome of humans [39Van Essen D.C. et al.The WU–Minn Human Connectome Project: an overview.Neuroimage. 2013; 80: 62-79Crossref PubMed Scopus (2780) Google Scholar]. These imaging techniques, however, produce relatively coarse grained and noisy maps. In contrast to the structural connectome, the functional connectome is inherently less precise and more difficult to define [40Horwitz B. The elusive concept of brain connectivity.Neuroimage. 2003; 19: 466-470Crossref PubMed Scopus (599) Google Scholar]. Functional interactions most meaningfully reflect directed causal relationships, but most present methods are only able to infer these relationships indirectly, through observation of undirected correlations. Furthermore, the possibility of such interactions to occur on multiple temporal scales places an additional challenge on the detection of frequency-specific interactions [3Alivisatos A.P. et al.The Brain Activity Map Project and the challenge of functional connectomics.Neuron. 2012; 74: 970-974Abstract Full Text Full Text PDF PubMed Scopus (379) Google Scholar]. Models of functional connectomes may differ on the spatial resolution, temporal resolution, and spatial extent of these maps. At the microscale, models of the functional connectome are based on extracellular recordings and optical calcium-imaging techniques, which allow the mapping of dynamical interactions between small groups of individual neurons of animals [41Lutcke H. et al.Steady or changing? Long-term monitoring of neuronal population activity.Trends Neurosci. 2013; 36: 375-384Abstract Full Text Full Text PDF PubMed Scopus (70) Google Scholar, 42Tye K.M. Deisseroth K. Optogenetic investigation of neural circuits underlying brain disease in animal models.Nat. Rev. Neurosci. 2012; 13: 251-266Crossref PubMed Scopus (566) Google Scholar] and between cultured neurons developed from stem cells or fibroblasts of patients with psychiatric disorders [43Brennand K.J. et al.Modeling psychiatric disorders at the cellular and network levels.Mol. Psychiatry. 2012; 17: 1239-1253Crossref PubMed Scopus (89) Google Scholar]. At the macroscale, functional MRI allows the study of interactions between brain regions in living animals [44Hutchison R.M. Everling S. Monkey in the middle: why nonhuman primates are needed to bridge the gap in resting-state investigations.Front. Neuroanat. 2012; 6: 29Crossref PubMed Scopus (98) Google Scholar] or humans [45Biswal B.B. et al.Toward discovery science of human brain function.Proc. Natl. Acad. Sci. U.S.A. 2010; 107: 4734-4739Crossref PubMed Scopus (2185) Google Scholar] and neurophysiological recordings allow the study of high-frequency interactions between broader (less well resolved) areas of brain tissue [46Siegel M. et al.Spectral fingerprints of large-scale neuronal interactions.Nat. Rev. Neurosci. 2012; 13: 121-134Crossref PubMed Scopus (846) Google Scholar]. Analyses of empirical models of the connectome aim to describe the organization of these models by characterizing local and global network topology patterns. Simple connectome models are networks of nodes (neurons, neuronal ensembles, or brain regions) and edges or links (synapses, projections, tracts, correlational, or causal interactions) [47Rubinov M. Sporns O. Complex network measures of brain connectivity: uses and interpretations.Neuroimage. 2010; 52: 1059-1069Crossref PubMed Scopus (7135) Google Scholar]; more sophisticated models may additionally include types of nodes and links [48Qian J. et al.Colored motifs reveal computational building blocks in the C. elegans brain.PLoS ONE. 2011; 6: e17013Crossref PubMed Scopus (28) Google Scholar, 49Rubinov M. Sporns O. Weight-conserving characterization of complex functional brain networks.Neuroimage. 2011; 56: 2068-2079Crossref PubMed Scopus (576) Google Scholar] as well as directionality and weight of links [50Hagmann P. et al.Mapping the structural core of human cerebral cortex.PLoS Biol. 2008; 6: e159Crossref PubMed Scopus (3036) Google Scholar, 51Varshney L.R. et al.Structural properties of the Caenorhabditis elegans neuronal network.PLoS Comput. Biol. 2011; 7: e1001066Crossref PubMed Scopus (547) Google Scholar]. Models of the connectome have many nodes and links and are difficult to characterize qualitatively. Investigators seek general rules for the global organization of these models. Early rules for brain-network organization postulated simple principles of this organization. For instance, Ramón y Cajal's principle of wiring economy proposes that anatomical connections are primarily governed by minimization of neuronal wiring [52Cajal S.R. Histology of the Nervous System of Man and Vertebrates. Oxford University Press, 1995Google Scholar, 53Bullmore E. Sporns O. The economy of brain network organization.Nat. Rev. Neurosci. 2012; 13: 336-349Crossref PubMed Scopus (2004) Google Scholar], whereas Peters' rule proposes that anatomical connections nonspecifically form in proportion to synaptic density [54Braitenberg V. Schüz A. Peters' rule and White's exceptions.Cortex: Statistics and Geometry of Neuronal Connectivity. Springer, 1998: 99-101Crossref Google Scholar, 55Reid R.C. From functional architecture to functional connectomics.Neuron. 2012; 75: 209-217Abstract Full Text Full Text PDF PubMed Scopus (52) Google Scholar]. Together, these rules propose a random organization of synaptic connections within spatially constrained clusters. More recently, complex-network analysis tools [56Newman M. The structure and function of complex networks.SIAM Rev. 2003; 45: 167-256Crossref Scopus (13431) Google Scholar] have allowed to probe the whole-brain organization of connectome models. One of the main insights of these analyses is the finding of simultaneous and partial reconciliation of the principles of economical and nonspecific (random) wiring [8Bullmore E. Sporns O. Complex brain networks: graph theoretical analysis of structural and functional systems.Nat. Rev. Neurosci. 2009; 10: 186-198Crossref PubMed Scopus (7508) Google Scholar, 53Bullmore E. Sporns O. The economy of brain network organization.Nat. Rev. Neurosci. 2012; 13: 336-349Crossref PubMed Scopus (2004) Google Scholar]. These analyses collectively demonstrate a clustered network organization with a substantial number of long-range connections and the presence of prominent central neural elements or hubs. We focus on whole-brain principles of connectome organization, but note that the description of specific nodes or connections [57Briggman K.L. et al.Wiring specificity in the direction-selectivity circuit of the retina.Nature. 2011; 471: 183-188Crossref PubMed Scopus (590) Google Scholar] or specific network clusters [58Damoiseaux J.S. et al.Consistent resting-state networks across healthy subjects.Proc. Natl. Acad. Sci. U.S.A. 2006; 103: 13848-13853Crossref PubMed Scopus (3298) Google Scholar] of connectome models represent important complementary analysis approaches. Although the focus on whole-brain organization reflects our research interests, we note that psychiatric disorders manifest abnormalities in other models of the connectome [43Brennand K.J. et al.Modeling psychiatric disorders at the cellular and network levels.Mol. Psychiatry. 2012; 17: 1239-1253Crossref PubMed Scopus (89) Google Scholar] or with other analyses, such as regional-connectivity mapping methods [59Zhang D. Raichle M.E. Disease and the brain's dark energy.Nat. Rev. Neurol. 2010; 6: 15-28Crossref PubMed Scopus (698) Google Scholar]. These models and analyses are equally associated with all the above-described conceptual and methodological limitations of pathoconnectomics. In this section we provide an overview of recently reported whole-brain abnormalities of anatomical and functional MRI-based connectome models of schizophrenia and autism. The focus on schizophrenia and autism reflects the weight of the current literature; there are considerably fewer and less conclusive results of whole-brain connectome organization for other psychiatric disorders, such as major depression, bipolar disorder, and attention-deficit/hyperactivity disorder. MRI-based structural models of the connectome infer the presence of white-matter fibers from patterns of uneven (anisotropic) water diffusion, whereas MRI-based functional models of the connectome infer the presence of functional interactions from correlations of fluctuations in the blood-oxygen-level dependent signal, an indirect measure of neuronal activity. Important issues for construction of MRI-based connectome models include the adequate removal of noise such as head-motion artifact, spatial normalization of images to a common template for between-group comparisons, accurate definitions of brain regions as nodes, and robust and reproducible inference of structural or functional links [38Craddock R.C. et al.Imaging human connectomes at the macroscale.Nat. Methods. 2013; 10: 524-539Crossref PubMed Scopus (278) Google Scholar]. The lack of a standardized analysis pipeline is an important additional problem which makes it difficult to compare results between different studies [107Carp J. The secret lives of experiments: methods reporting in the fMRI literature.NeuroImage. 2012; 63: 289-300Crossref PubMed Scopus (279) Google Scholar]. Network analyses of whole-brain organization often make inferences about brain activity, such as the propensity or presence of segregation and integration of brain dynamics, from properties of whole-brain organization. Such inferences are based on measures of network topology such as the 'small-world' property, the simultaneous presence of clustered and distributed network topology [60Watts D.J. Strogatz S.H. Collective dynamics of 'small-world' networks.Nature. 1998; 393: 440-442Crossref PubMed Scopus (31091) Google Scholar]. In structural networks, such analyses reflect the potential for properties of brain dynamics to emerge on a structural substrate, but are based on generic assumptions such as the propagation of information along shortest paths. In functional networks, the situation is even less straightforward because links represent the presence of functional interactions and hence cannot measure the potential for dynamics to emerge on these interactions [49Rubinov M. Sporns O. Weight-conserving characterization of complex functional brain networks.Neuroimage. 2011; 56: 2068-2079Crossref PubMed Scopus (576) Google Scholar]. The relevance of such interpretations is hence unclear and we omit these interpretations below. The total number or total 'weight' of links is a simple whole-brain analysis of the connectome. This analysis describes the potential, presence, or extent of whole-brain network interactions and is relatively easy to interpret. The total number and weight of links additionally influences higher-order measures of network organization, such as clustered and distributed topology [61van Wijk B.C.M. et al.Comparing brain networks of different size and connectivity density using graph theory.PLoS ONE. 2010; 5: e13701Crossref PubMed Scopus (806) Google Scholar]. Early studies of regional connectivity in schizophrenia show redu
Referência(s)