Elsevier

NeuroImage

Volume 196, 1 August 2019, Pages 248-260
NeuroImage

Determinants of structural segregation and patterning in the human cortex

https://doi.org/10.1016/j.neuroimage.2019.04.031Get rights and content

Highlights

  • The cortical surface can be subdivided into 13 independent regions (“communities”).

  • Thirty centers of low inter-subject variability were identified in deep sulci.

  • The structural pattern within a community is governed typically by two centers.

  • Their variability showed a strong positive correlation with the known time points at which they appear in fetal development.

Abstract

This study aimed at uncovering mechanisms that govern the spatio-temporal patterning of the human cortex and its structural variability, and drawing links between fetal brain development and variability in adult brains. A data-driven analytic approach based on structural MR images revealed the following findings: (1) The cortical surface can be subdivided into 13 independent regions (“communities”) based on macroscopic features. (2) Thirty centers of low inter-subject variability were found in major sulci on the cortical surface. Their variability showed a strong positive correlation with the known time points at which they appear in fetal development. Centers forming early induce a higher inter-subject regularity in a larger local vicinity, while those forming later result in smaller regions of higher variability. (3) The layout of sulcal and gyral patterns within a community is governed typically by two centers. Depending on the relative variability of each center, communities can be classified into structural sub-types. (4) Sub-types across ipsi-lateral communities are independent, but associated with the sub-type of the same community on the contra-lateral side. Results shown here integrate well with current knowledge about macroscopic, microscopic, and genetic determinants of brain development.

Introduction

The quest to understand the relationship between brain and behavior has typically included an intermediate step in the form of brain parcellation. Indeed, the individual variability of macro-structural features of the cerebral cortex has puzzled neuro-anatomists for two centuries (Gratiolet, 1854; Eberstaller, 1890; Cunningham, 1892; Duvernoy, 1991), who, from visual observation, developed a common ontology (Swanson, 2015). Considerable training is required for a human observer to recognize cortical structures. Difficulties arise because even major sulci and gyri show a remarkable structural variability within and between subjects (Ono et al., 1990). Secondary features may be prevalent in some individuals only, and there is an abundance of literature describing detailed regional variation (e.g., Tomaiuolo et al., 1999; Chiavaras and Petrides, 2000; Kringelbach and Rolls, 2004; Germann et al., 2005; Kahnt et al., 2012). However, the definition of regions under study and criteria for sub-type classification are often phenomenological and ad-hoc. An over-arching “theory of cortical variability” trying to assess and interpret the relationship between common and individual features of the human cortex is currently missing.

Existing brain atlas schemes used in anatomical and functional MRI studies (e.g., Tzourio-Mazoyer et al., 2002; Desikan et al., 2006; Oishi et al., 2008; Glasser et al., 2016) are based on the implicit assumption that all brains can be sufficiently represented as a common mean, without assessing the validity of this assumption. The positional error incurred from mapping an individual brain into a normative space is generally well above the current resolution limit of modern neuroimaging methods. No current approach allows referencing the individual variability found on the cortical surface at the imaging level of detail, limiting the advance of knowledge about individual determinants in translational research and clinical practice.

In the present work, rather than focusing on commonalities, we aimed to elucidate mechanisms that govern cortical variability. We applied a data-driven, quantitative approach for assessing individual cortical variability, based on the analysis of anatomical MR brain imaging data acquired in a large cohort of healthy adults. Without including anatomical knowledge, our unbiased approach used machine-learning methods to reveal processes that segregate cortical space while allowing individual variations. As a first step, we introduce background and terminology used in this study:

Sulcal roots: Neuroanatomical studies of fetal development have established locations and time points at which sulci first appear on the cortical surface (Cunningham, 1892; Chi et al., 1977; Nishikuni and Ribas, 2012). Regis et al. (1995, 2005) introduced the term “sulcal roots” for these initial locations, and was one of the first to suggest the analysis of structural variations in terms of their number and relative configuration. More recently, analyses of in utero MRI data sets (Dubois et al. 2008, 2014; Hu et al., 2011; Clouchoux et al., 2012; Habas et al., 2012) have provided hints that sulcal roots are retained as the locally deepest points within sulci (“pits”). Studies in infants (Meng et al., 2014) and adults (Lohmann et al., 2008; Im et al., 2010; le Guen et al., 2018a) have shown that pits can serve as anatomical landmarks that have a low inter-subject variability (for a recent review, refer to: Im and Grant, 2018).

Basins: We introduced the concept of sulcal basins as a richer representation of cortical features than points (Yang and Kruggel, 2008; Kruggel, 2018). A basin corresponds to a cortical patch that is centered around a locally deepest point and includes the neighboring sulcal area up to the gyral crowns (Fig. 1). Basins are automatically segmented on a surface representing the gray/white matter (GM/WM) interface, using as criteria, geodesic depth and surface curvature. In this way, basins provide a complete segmentation of the cortical surface.

Variability: In our previous publication (Kruggel, 2018), we demonstrated how basin segmentations can be used to quantify inter-individual structural variability. In this context, cortical variability is a point-wise measure that corresponds to a weighted sum of the probabilities of finding specific basins at that location across a group of subjects. If the same basin is found at a specific location across all subjects, the variability is zero. Higher values indicate a higher structural variability (Fig. 2). We demonstrated that neighboring basins cluster into cortical communities: basins overlap within a community but not across communities when compared across subjects. Thirteen communities were derived by automated clustering (Fig. 3).

In this work, we assess biological mechanisms that govern cortical variability. We will demonstrate that: (1) The structural pattern within the 13 communities described before (Kruggel, 2018) is governed by 30 centers of low inter-subject variability (CLV) located in major sulci. (2) Their variability shows a strong positive correlation with the known time points at which they appear in fetal development. Deep, low variable centers are formed early in development. (3) The layout of sulcal and gyral patterns within a community is governed by typically two centers. Deep, low variable centers induce a higher regularity in a larger local vicinity. Depending on the relative variability of each center, communities can be classified into structural sub-types. (4) Sub-types across ipsi-lateral communities are independent, but associated with the sub-type of the same community on the contra-lateral side. Community sub-types offer a “pattern library” that is sufficiently large to represent individual variability.

Section snippets

Subjects and imaging data

This work included imaging data of the 1113 subjects in the “1200 Subjects Release” of the Human Connectome Project released in March 2017. The cohort consisted of 606 females and 507 males in the age range of 22–37 years (mean: 28.7 years). T1-and T2-weighted structural MR images were used in this study. For detailed acquisition information, refer to the release document (Human Connectome Project, 2017).

Segmentation of cortical features

The analytical procedures described in this section were extended from our previous work (

Results

On the basis of the analytical framework detailed below, we examined the relevance of cortical communities in terms of their relationship to brain development, and their importance for segregating the developing cortex into structural sub-units. First, we studied the relationship between communities and their centers of low variability (CLV). Second, we assessed the relationship of geodesic depth and variability of these centers, and linked their variability to their temporal sequence in fetal

Discussion

This study aimed at understanding the spatio-temporal determinants of structural segregation and patterning on the human cerebral cortex. We assessed the spatial relationship between regions common to all brains (“communities”) and their individual structural variants (“sub-types”), and provided clues about their temporal allocation.

Our principal findings can be summarized as follows: (1) The cortical surface can be subdivided into 13 communities based on macroscopic features. These communities

Acknowledgment

Authors thank Dr. Robert Hunt (Depts. of Anatomy and Neurobiology, UC Irvine) for insightful comments on the manuscript.

References (79)

  • F. Kruggel

    The macro-structural variability of the human neocortex

    Neuroimage

    (2018)
  • Y. le Guen et al.

    The chaotic morphology of the left superior temporal sulcus is genetically constrained

    Neuroimage

    (2018)
  • Y. Meng et al.

    Spatial distribution and longitudinal development of deep cortical sulcal landmarks in infants

    Neuroimage

    (2014)
  • K. Oishi et al.

    Human brain white matter atlas: identification and assignment of common anatomical structures in superficial white matter

    Neuroimage

    (2008)
  • J.A. Scott et al.

    Growth trajectories of the human fetal brain tissues estimated from 3D reconstructed in utero MRI

    Int. J. Dev. Neurosci.

    (2011)
  • N. Tzourio-Mazoyer et al.

    Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain

    Neuroimage

    (2002)
  • H.B.M. Uylings et al.

    3D Cytoarchitectonic parcellation of human orbitofrontal cortex. Correlation with postmortem MRI

    Psychiatr. Res. Neuroimaging

    (2010)
  • F. Yang et al.

    Automatic segmentation of human brain sulci

    Med. Image Anal.

    (2008)
  • A. Anwander et al.

    Connectivity-based parcellation of Broca's area

    Cerebr. Cortex

    (2007)
  • H.J. Bandelt et al.

    Metric graph theory and geometry: a survey

    Contemp. Math.

    (2008)
  • P.V. Bayly et al.

    A cortical folding model in- corporating stress-dependent growth explains gyral wavelengths and stress patterns in the developing brain

    Phys. Biol.

    (2013)
  • P. Billie

    A survey on tree edit distance and related problems

    Theor. Comput. Sci.

    (2005)
  • V. Borrell

    How cells fold the cerebral cortex

    J. Neurosci.

    (2018)
  • R. Campigotto et al.

    A generalized and adaptive method for community detection

    (2014)
  • J.G. Chi et al.

    Gyral development of the human brain

    Ann. Neurol.

    (1977)
  • M.M. Chiavaras et al.

    Orbitofrontal sulci of the human and macaque monkey brain

    J. Comp. Neurol.

    (2000)
  • C. Clouchoux et al.

    Quantitative in vivo MRI measurement of cortical development in the fetus

    Brain Struct. Funct.

    (2012)
  • D.J. Cunningham

    Contribution to the Surface Anatomy of the Cerebral Hemispheres

    (1892)
  • C. de Juan Romero et al.

    Discrete domains of gene expression in germinal layers distinguish the development of gyrencephaly

    EMBO J.

    (2015)
  • R.S. Desikan et al.

    An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest

    Neuroimage

    (2006)
  • D. Duan et al.

    Exploring folding patterns of infant cerebral cortex based on multi-view curvature features: methods and applications

    Neuroimage

    (2018)
  • J. Dubois et al.

    Primary cortical folding in the human newborn: an early marker of later functional development

    Brain

    (2008)
  • J. Dubois et al.

    The dynamics of cortical folding waves and prematurity-related deviations revealed by spatial and spectral analysis of gyrification

    Neuroimage

    (2019)
  • H.M. Duvernoy

    The Human Brain

    (1991)
  • O. Eberstaller

    Das Stirnhirn: Ein Beitrag zur Anatomie der Oberfläche des Grosshirns

    (1890)
  • V. Fernandez et al.

    Cerebral cortex expansion and folding: what have we learned?

    EMBO J.

    (2016)
  • M. Ferrer et al.

    Graph edit distance - theory, algorithms, and applications

  • A.M. Fish et al.

    Influences of brain size, sex, and sex chromosome complement on the architecture of human cortical folding

    Cerebr. Cortex

    (2017)
  • S. Fortunato

    Community detection in graphs

    Phys. Rep.

    (2008)
  • Cited by (0)

    View full text