Topographic organization of the cerebral cortex and brain cartography

Abstract

One of the most specific but also challenging properties of the brain is its topographic organization into distinct modules or cortical areas. In this paper, we first review the concept of topographic organization and its historical development. Next, we provide a critical discussion of the current definition of what constitutes a cortical area, why the concept has been so central to the field of neuroimaging and the challenges that arise from this view. A key aspect in this discussion is the issue of spatial scale and hierarchy in the brain. Focusing on in-vivo brain parcellation as a rapidly expanding field of research, we highlight potential limitations of the classical concept of cortical areas in the context of multi-modal parcellation and propose a revised interpretation of cortical areas building on the concept of neurobiological atoms that may be aggregated into larger units within and across modalities. We conclude by presenting an outlook on the implication of this revised concept for future mapping studies and raise some open questions in the context of brain parcellation.

Figure 1

Autoradiographic visualization of the distribution of binding sites for three different neurotransmitter receptors as well as a corresponding cell body stained section. Cortical borders within the early visual cortex as evident by each feature are marked in the image and areas are labeled in the cytoarchitectonic section. Two important aspects become evident from this histological data. First, not every feature or modality reveals all cortical borders. Second, in case that multiple features indicate a given border, the locations of these borders are usually in very good spatial agreement

Figure 2

Multi-modal features of cortical differentiation exemplified on a location within Area 32s of the cingulate cortex (Palomero-Gallagher et al., 2008). The region of interest is displayed in the top left image in green. In clockwise order, the displayed features for this region are: Receptor- and cytoarchitecture. Functional characterization based on task-activation data via the BrainMap database, illustrating which kind of tasks elicit activation in that region. Group-average resting-state functional connectivity (positive connectivity is shown in red, negative connectivity in blue). Structural covariance in local grey matter volume as estimated across ~200 subjects based on standard VBM processing. Anatomical connectivity as estimated via probabilistic tractography averaged over 100 subjects

Figure 3

Schematic summary of the proposed relationship between uni- and multi-modal definitions of cortical areas in a hierarchically organized system. For simplicity, cortical location is presented along a single dimension. The top seven rows represent cortical borders as defined by individual modalities. The height of the vertical bars represent the conspicuousness of putative borders, i.e., the degree of difference between the modules on either side in this modality. Applying an (in the end arbitrary) threshold for the acknowledgement or rejection of these borders yields uni-modal parcellations as indicated by the black and grey shading of the horizontal bar (for simplicity, only two colors were used, this does not indicate that the different black segments represent the same area). The bottom plot shows the superimposition of the uni-modal borders and illustrates that borders usually coincide but may not be revealed by all features. Rather, some borders will only be reflected in a few modalities, representing highly similar modules, whereas borders between what would traditionally be considered cortical areas would be reflected by more consistent (between features) and stronger (within features) differences

Figure 4

Figure X. Cortical parcellations based on histology or in-vivo brain imaging. (A) Jueliech and (B) Allen Brain Institute multi-modal cortical areas defined using cyto- and chemo-architectonics (Amunts et al., 2015; Ding et al., 2016). (C) Cortical areas defined by fMRI visuotopic mapping in an individual subject (Wang et al., 2015). (D, E, F) Resting-state functional connectivity parcellations by (D) Shen et al. (2013), (E) Gordon et al. (2016), and (F) Schaefer et al. (2016). (G) Diffusion MRI connectivity parcellation (Fan et al., in press). (H) Multi-modal parcellation with meta-analytic connectivity mapping (MACM), resting-state fMRI and diffusion connectivity, shown here for MACM (Genon et al., in press). (H) Parcellation of right dorsal premotor region with MACM, resting-state fMRI and diffusion connectivity, shown here for MACM (Genon et al., in press). (I) Multimodal parcellation using resting-state functional connectivity, relative myelin mapping, cortical thickness and task fMRI (Glasser et al., 2016). Figures courtesy of authors

Figure 5

(A) Map of functional connectivity gradient strength and (B) resulting parcellation (Gordon et al., 2016). Functional connectivity gradient strength measures how fast the functional connectivity is changing over the cortical surface at a spatial location. A higher value indicates that the functional connectivity maps are rapidly changing at that spatial location. Green asterisk indicates very weak gradient between adjacent green and yellow parcels. In contrast, white asterisks indicates strong gradient between adjacent green and orange parcels. Figures courtesy of EMG