A deformation-based approach for characterizing brain asymmetries at different spatial scales of resolution

Abstract

Background.-Structural cerebral asymmetries are hypothesized to provide an architectural foundation for functional asymmetries and behavioral lateralities. Studies of structural asymmetries typically focus on gray matter measures that are influenced by gross deformation fields used for normalization, and thus characterize a combination of different morphologic influences on structural asymmetries. New Method.-A deformation-based morphometry approach was developed to characterize structural asymmetries at different spatial scales of resolution, which can provide relatively more specific inference about the morphologic reason(s) for structural asymmetries, using a dataset of 347 typically developing children (7.00-12.92 years). Results.-Significant structural asymmetries were observed for a larger lobar spatial scale (e.g., frontal petalia) and for a smaller gyral/sulcal spatial scale of resolution (e.g., marginal sulcus). Total intracranial volume was significantly associated with asymmetries at the larger spatial scale of normalization, while age was significantly associated with asymmetries at the smaller scale of normalization. There were no significant anti-or fluctuating asymmetry effects based on Hartigan Dip Tests and Bonnett Tests, respectively. Comparison with Existing Method(s).-While spatially similar asymmetries were observed in both gray matter and deformation field data (e.g., medial planum temporale / Heschl's gyrus), the deformation approach characterizes asymmetries based on three iterations of successively smaller scales of normalization. Conclusions.-Structural asymmetries can be identified in normalization deformations with a procedure that is tailored for sensitivity to structures at different spatial scales of resolution where there may be different mechanisms for the expression of asymmetry.

The Topology of Pediatric Structural Asymmetries in Language-Related Cortex

“…Here those structures represented hemispheric differences in the amount of volumetric displacement needed to warp the images to a symmetrical template. At the voxel level, these asymmetry data exhibit the same type of gray matter and white matter asymmetries that have been described previously [8] and appear to yield similar results to surface area asymmetries [12].…”

“…The Jacobian asymmetry results presented later can be interpreted in the context of other voxel-based studies of gray and white matter asymmetries where Jacobian modulation of the gray or white matter voxel values was performed. That is, the voxel-level asymmetries with this Jacobian or deformation-based approach are very similar to the voxel-level asymmetries involving specific tissue types [8]. Given the similar pattern of results between voxel-based gray matter volume asymmetries and surface area asymmetries [9,12], the current volumetric deformation approach was also expected to yield results that are similar to surface area asymmetry results.…”

“…Specifically, 1) ten normalization iterations were initially performed to create the optimal template where the mean of the images from the preceding step served as the normalization target for the next iteration [normalization parameters-iterations 1-9 (30 × 90 × 20 × 12); iteration 10 (30 × 90 × 30 × 20); across iterations: cross-correlation metric (mm radius = 4.00); SyN (2, 1, 1); Gaussian regularization (4.00)]. The optimal symmetrical template was then 2) used as a target for a series of three normalization iterations that varied in the magnitude of volumetric displacement, as described in [8] [normalization parameters-iteration 1 (50 × 1 × 1): cross-correlation metric (mm radius = 4.00), SyN (1.00), Gaussian regularization (3.00); iteration 2 (1 × 50 × 1): cross-correlation metric (mm radius = 2.00), SyN (0.75), Gaussian regularization (2.00); iteration 3 (1 × 1 × 40): cross-correlation metric (mm radius = 1.00), SyN (0.50), Gaussian regularization (1.00)].…”

“…A defining feature of the human brain is its asymmetric structure, including in language-related brain regions that exhibit leftward asymmetric activation during language tasks [5][6][7]. Leftward anterior insula asymmetry and rightward superior temporal sulcus asymmetry, for example, have been consistently observed across surface area and volumetric measurement methods where univariate voxel-level analyses were performed in pediatric [8,9] and adult samples [10][11][12][13]. While variation in insular Symmetry 2020, 12, 1809 2 of 17 asymmetry has been related to estimates of language dominance and asymmetric activation in at least three studies where volumetric asymmetries were examined [14][15][16], measures of other language-related regions have been inconsistently associated with language laterality [14,[16][17][18][19][20].…”

“…Sex has been related to structural asymmetries in multiple voxel-based studies [10,13,21], although non-significant results have been reported [8,11], perhaps because sex differences in brain size may explain locally specific sex effects in some samples [22]. The magnitude of sex differences may also depend on age.…”