The individuality of shape asymmetries of the human cerebral cortex

Abstract

Asymmetries of the cerebral cortex are found across diverse phyla and occur at multiple spatial scales, from the global morphology of the two hemispheres to fine-grained sulcal and gyral anatomy. Cortical asymmetries are particularly pronounced in humans and have important implications for brain function and disease. However, most studies conflate asymmetries in regional size and shape and only consider effects at fairly fine-grained resolutions, such as image voxels or a priori regions-of-interest. As such, whether asymmetries are preferentially expressed at specific spatial resolutions over others remains unclear. Here, we apply a spectral shape analysis to magnetic resonance imaging (MRI) data from three independent samples to derive a multiscale description of asymmetries in cortical shape that is distinct from regional size-related descriptors such as volume or thickness. We show that our proposed shape asymmetry signature (SAS) is a highly individualized feature, akin to a cortical fingerprint, that more accurately identifies individuals than size-based descriptors or measures of inter-regional functional coupling of brain activity. Notably, individual identifiability is optimal at coarse spatial scales (~37 mm wavelength), and shape asymmetries at specific scales show differences between males and females but are unrelated to the handedness. We further demonstrate that while unihemispheric shape descriptors show significant heritability at coarse scales, the SAS are determined primarily by unique environmental effects across all scales considered. Our findings thus identify coarse-scale scale asymmetries of cortical shape as being highly unique to individuals, sexually dimorphic, and largely driven by stochastic environmental influences.

Atypical language network lateralization is reflected throughout the macroscale functional organization of cortex

“…The parallel and interdigitated organization of cortical networks suggests that the language system may impinge upon, and be influenced by, putatively distinct yet spatially adjacent networks 88 . Extending upon recent evidence indicating that asymmetries in regional brain anatomy are reflected throughout the entire brain 89,90 , our analyses revealed a relationship between the lateralizations of language functions and the sweeping functional gradients that capture the topography of large-scale networks across cortex. Of note, across all three gradients the observed differences between the typical and atypical groups were preferential to association networks.…”

Mode-based morphometry: A multiscale approach to mapping human neuroanatomy

“…(2) on the sphere, degenerate solutions exist, such that certain eigenmodes will have the same number of nodal lines and wavelengths, and they can be aggregated into an eigengroup l, similar to the angular momentum number in quantum physics. Each eigengroup comprises 2l `1 eigenmodes, and its wavelength is [46,41] wavelength " 2πR s a lpl `1q ,…”

“…where R s is the radius of the sphere (see Table S1 for an explicit list of wavelengths and eigenmode membership for each eigengroup on a sphere of R s " 67 mm [41,15]). Therefore, the eigenmodes of the cortical surface belonging to a given eigengroup have approximately the same spatial scale or wavelength given by Eq.…”

“…Note that individual-specific eigenmodes can also be derived from individual surfaces. However, small differences in the geometry of the cortical surface can alter the spatial patterns of the eigenmodes, which makes comparison between individual-specific modes challenging, particularly for short wavelengths or more fine-grained modes [41]. Moreover, prior work indicates that relying on a single, canonical basis set of modes derived from a group template leads to a negligible difference in reconstructions of empirical data when compared to using individual-specific modes [15], suggesting that the group-derived modes provide a reasonable approximation for present purposes.…”

“…Indeed, recent work using geometric eigenmodes suggests that virtually all task-evoked activation maps obtained with functional MRI result from excitations of lowfrequency modes spanning wavelengths ą60 mm [15]. Other work indicates that individual differences in brain anatomy are most salient at relatively coarse scales, spanning wavelengths ě37 mm [41]. Such findings challenge classical approaches to brain mapping that focus on identifying focal effects in isolated brain regions and underscore the benefits of a spatially-informed, multiscale approach.…”