Elsevier

NeuroImage

Volume 69, 1 April 2013, Pages 51-61
NeuroImage

The connectivity of functional cores reveals different degrees of segregation and integration in the brain at rest

https://doi.org/10.1016/j.neuroimage.2012.11.051Get rights and content

Abstract

The principles of functional specialization and integration in the resting brain are implemented in a complex system of specialized networks that share some degree of interaction. Recent studies have identified wider functional modules compared to previously defined networks and reported a small-world architecture of brain activity in which central nodes balance the pressure to evolve segregated pathways with the integration of local systems. The accurate identification of such central nodes is crucial but might be challenging for several reasons, e.g. inter-subject variability and physiological/pathological network plasticity, and recent works reported partially inconsistent results concerning the properties of these cortical hubs. Here, we applied a whole-brain data-driven approach to extract cortical functional cores and examined their connectivity from a resting state fMRI experiment on healthy subjects. Two main statistically significant cores, centered on the posterior cingulate cortex and the supplementary motor area, were extracted and their functional connectivity maps, thresholded at three statistical levels, revealed the presence of two complex systems. One system is consistent with the default mode network (DMN) and gradually connects to visual regions, the other centered on motor regions and gradually connects to more sensory-specific portions of cortex. These two large scale networks eventually converged to regions belonging to the medial aspect of the DMN, potentially allowing inter-network interactions.

Highlights

► We applied a data-driven approach to extract cores of connectivity at rest. ► Two main statistically significant cores: PCC and SMA, were extracted. ► Two highly segregated networks are identified: ► The default net gradually connecting to visual and prefrontal regions ► The motor net gradually connecting to more sensory-specific portions of cortex

Introduction

The existence of a structured pattern of neuronal activity generated in the brain in the absence of an explicit task (i.e. in the resting state) is now a well reported phenomenon. Multiple techniques (e.g. fMRI, MEG) showed that spontaneous, slow (< 0.1 Hz) fluctuations of cerebral activity are temporally coherent within widely distributed functional networks that closely resemble those evoked by sensory, motor, and cognitive paradigms (Biswal et al., 1997, de Pasquale et al., 2010, de Pasquale et al., 2012, Fox et al., 2007, He et al., 2008, Nir et al., 2008). Therefore, the two fundamental properties of functional segregation and dynamic integration (Friston, 2002), which have characterized previous models of brain activity focusing at the local scale, i.e. related to specific regions of interest, need now to be integrated in a network model.

Initially, resting state networks (RSN) emerged as segregated functional systems of highly coupled nodes (Biswal et al., 1997, Fox et al., 2005). The existence of an internal network architecture was later hypothesized in which distinct nodes played different functional roles, a view that has received experimental support by structural and functional connectivity studies (Andrews-Hanna et al., 2007, de Pasquale et al., 2010, He et al., 2007, Iturria-Medina et al., 2008, Tomasi and Volkow, 2010). Finally, recent studies have provided evidence of a consistent interaction across networks over time (de Pasquale et al., 2012, Deshpande et al., 2009, Deshpande et al., 2011). Therefore, the two original principles of functional specialization and integration seem to be implemented in a complex system of internally structured, and functionally specialized, networks that share some degree of interaction. Consequently, the original distinction in separate resting state networks does not appear to be such a clear cut. As a matter of fact, recent studies based on graph theory have shown the emergence of wider functional modules compared to the originally identified networks, in which components from different networks are grouped together (e.g. motor and auditory areas; see (He et al., 2009, Meunier et al., 2009a, Meunier et al., 2009b)). In this context, data from different imaging modalities suggested the small-world architecture (Achard et al., 2006, Bullmore and Sporns, 2009, Stam, 2004) as an efficient model of segregation and integration in the brain. According to this model, central nodes, characterized by many local connections and a few distant ones, serve to balance the pressure to evolve segregated pathways with the integration of local networks by minimizing the cost of wiring and metabolism (Bassett et al., 2006).

The accurate localization of such nodes is crucial to understanding the brain functional architecture in healthy aging (Alalade et al., 2011, Galvin et al., 2011, Oghabian et al., 2010, Westlye et al., 2011) as well as in disease (Buckner et al., 2009, Greicius, 2008, Li et al., 2011, Wu et al., 2009). In this work we adopted the term ‘core’ to identify a node characterized by a high number of connections in the whole brain, while we used the term ‘hub’ when referring to specific graph theory results (Guimera and Amaral, 2005, Meunier et al., 2010). The identification of functional cores may be challenging for several reasons. For example, the use of nodes obtained from independent studies may not be the optimal choice, due to both inter-subject variability and to the eventual presence of physiological (e.g. healthy aging) or pathological (e.g. recovery from a brain injury) network plasticity (Castellanos et al., 2011). Furthermore, recent studies based on graph theory analyses have reported interesting, but partially inconsistent, results concerning hub location. The observed discrepancies could be ascribed either to methodological differences (see for example (Buckner et al., 2009, Tomasi and Volkow, 2011a, Tomasi and Volkow, 2011b)) or to the fact that, due to the computational complexity, these studies are typically based on a limited set of regions of interest within the brain. To overcome these limitations, we have investigated here the architecture of connections involving functional brain cores by using a whole-brain data-driven approach. First, we identified cores based on their statistically significant connections within the brain and their reproducibility across subjects. Then, we proposed a multi-level organization of these cores based on a statistical criterion. Finally, we examined the spatial topography of connections involving the identified cores.

Section snippets

Experimental setup

Twenty healthy subjects (9 women and 11 men; mean age ± standard deviation: 30 ± 10 years) provided informed written consent and participated in this study, which was approved by the local ethics committee. The study comprised four consecutive resting state sessions. Subjects were told to stay still and relaxed and no particular instructions were given to attend or fixate a particular stimulus. This strategy was adopted to simulate the acquisition of data from patients whose clinical conditions may

Results

Table 1 shows the mean and standard deviation of the center of mass representing the extracted candidate cores showing the highest centrality, defined as the total number of significant connections, at the three different statistical α levels. Core location appears reproducible across subjects, since the range of the standard deviations is in the order of 10 mm and thus we labeled these cores according to existing fMRI literature (de Pasquale et al., 2010, de Pasquale et al., 2012).

Fig. 2

Control analyses

In the SI we present several control analyses in which we investigated the spatial structure of the obtained networks, the reproducibility across subjects and the influence of PESTICA on the core identification. First we examined the influence of the size of connected regions on the proposed measure of centrality. As a matter of fact, although we have already excluded close connections due to second order neighbors (corresponding to approx. 12 mm), there is still the possibility that a single

Discussion

In this work, we have applied a whole-brain data-driven approach to extract cortical cores of functional connectivity (FC) in single individuals at rest and examined their spatial pattern of connectivity. Two main statistically significant cores, centered on PCC and SMA, were extracted considering the key properties of centrality and consistency across subjects. The FC maps associated with these core regions were thresholded at three significance levels, revealing the presence of two highly

Acknowledgments

This work was partially supported by the Italian Ministry of Health [grant RF2008 n.31].

References (62)

  • D. Meunier et al.

    Age-related changes in modular organization of human brain functional networks

    Neuroimage

    (2009)
  • K. Murphy et al.

    The impact of global signal regression on resting state correlations: are anti-correlated networks introduced?

    Neuroimage

    (2009)
  • A.J. Schwarz et al.

    Negative edges and soft thresholding in complex network analysis of resting state functional connectivity data

    Neuroimage

    (2011)
  • C.J. Stam

    Functional connectivity patterns of human magnetoencephalographic recordings: a ‘small-world’ network?

    Neurosci. Lett.

    (2004)
  • D. Tomasi et al.

    Functional connectivity hubs in the human brain

    Neuroimage

    (2011)
  • T. Wu et al.

    Changes of functional connectivity of the motor network in the resting state in Parkinson's disease

    Neurosci. Lett.

    (2009)
  • J.M. Abolafia et al.

    Slow modulation of ongoing discharge in the auditory cortex during an interval-discrimination task

    Front. Integr. Neurosci.

    (2011)
  • S. Achard et al.

    A resilient, low-frequency, small-world human brain functional network with highly connected association cortical hubs

    J. Neurosci.

    (2006)
  • E. Alalade et al.

    Altered cerebellar–cerebral functional connectivity in geriatric depression

    PLoS One

    (2011)
  • D.S. Bassett et al.

    Adaptive reconfiguration of fractal small-world human brain functional networks

    Proc. Natl. Acad. Sci. U. S. A.

    (2006)
  • Y. Benjamini et al.

    Controlling the false discovery rate: a practical and powerful approach to multiple testing

    J. R. Stat. Soc. B

    (1995)
  • B. Biswal et al.

    Functional connectivity in the motor cortex of resting human brain using echo-planar MRI

    Magn. Reson. Med.

    (1995)
  • B.B. Biswal et al.

    Simultaneous assessment of flow and BOLD signals in resting-state functional connectivity maps

    NMR Biomed.

    (1997)
  • U. Braun et al.

    Test–retest reliability of resting-state connectivity network characteristics using fMRI and graph theoretical measures

    Neuroimage

    (2011)
  • R.L. Buckner et al.

    The brain's default network: anatomy, function, and relevance to disease

    Ann. N. Y. Acad. Sci.

    (2008)
  • R.L. Buckner et al.

    Cortical hubs revealed by intrinsic functional connectivity: mapping, assessment of stability, and relation to Alzheimer's disease

    J. Neurosci.

    (2009)
  • E. Bullmore et al.

    Complex brain networks: graph theoretical analysis of structural and functional systems

    Nat. Rev. Neurosci.

    (2009)
  • A.R. Carter et al.

    Resting interhemispheric functional magnetic resonance imaging connectivity predicts performance after stroke

    Ann. Neurol.

    (2010)
  • N.P. Castellanos et al.

    Alteration and reorganization of functional networks: a new perspective in brain injury study

    Front. Hum. Neurosci.

    (2011)
  • D. Cordes et al.

    Frequencies contributing to functional connectivity in the cerebral cortex in “resting-state” data

    AJNR Am. J. Neuroradiol.

    (2001)
  • F. de Pasquale et al.

    Temporal dynamics of spontaneous MEG activity in brain networks

    Proc. Natl. Acad. Sci. U. S. A.

    (2010)
  • Cited by (41)

    • Cortical cores in network dynamics

      2018, NeuroImage
      Citation Excerpt :

      First, we briefly discuss different measures that have been adopted to identify structural or functional hubs in the brain (Bassett and Sporns, 2017; van den Heuvel and Sporns, 2013b). This is important since differences among these measures, and the different metrics to estimate the spatiotemporal structure of connectivity typically lead to discrepancies in the literature on the precise localization of central regions, (Buckner et al., 2009; Bullmore and Sporns, 2012; Cole et al., 2010; de Pasquale et al., 2012; de Pasquale et al., 2013; Hagmann et al., 2008; Power et al., 2013; Tomasi and Volkow, 2011; van den Heuvel and Sporns, 2013b). Next, we introduce the concept of dynamic core, defined as a set of brain regions showing the most consistent dynamic centrality with the rest of the brain (de Pasquale et al., 2016; de Pasquale et al., 2013).

    • Focal cortical dysplasia alters electrophysiological cortical hubs in the resting-state

      2015, Clinical Neurophysiology
      Citation Excerpt :

      The prominent involvement of midline structures in the FCD patients could be related to functional reorganization in the resting-state, presumably due to cortical dysplasia. The PCIN is a well-known component of the default mode network (DMN) (Greicius et al., 2003; Fransson and Marrelec, 2008; de Pasquale et al., 2013; Li et al., 2013), and MCIN has been reported to indicate the potential core of the ‘self’ (Northoff and Bermpohl, 2004; Fransson and Marrelec, 2008; Kim, 2012). Because midline structures are associated with the DMN (de Pasquale et al., 2013), the presence of MCIN can be regarded as another representation of the default state of the brain at rest.

    • Chinese Chorales Dataset: A High-Quality Music Dataset for Score Generation

      2024, Communications in Computer and Information Science
    View all citing articles on Scopus
    View full text