Normative cerebral cortical thickness for human visual areas

Abstract

Studies of changes in cerebral neocortical thickness often rely on small control samples for comparison with specific populations with abnormal visual systems. We present a normative dataset for FreeSurfer-derived cortical thickness across 25 human visual areas derived from 960 participants in the Human Connectome Project. Cortical thickness varies systematically across visual areas, in broad agreement with canonical visual system hierarchies in the dorsal and ventral pathways. In addition, cortical thickness estimates show consistent within-subject variability and reliability. Importantly, cortical thickness estimates in visual areas are well described by a normal distribution, making them amenable to direct statistical comparison.

Keywords: Cortical thickness; Extrastriate; FreeSurfer; Human connectome project; MRI; V1; Visual cortex.

Fig. 1

Spatial variability of cortical thickness across visual areas on the inflated normalized cortical surface. (A) Location of visual areas according to their maximum intensity projections in the (Wang et al., 2015) retinotopic atlas, and (B) between-subject mean curvature-corrected cortical thickness for 960 subjects of the HCP dataset. A positive gradient of increasing cortical thickness aligns with visual area cortical hierarchy along both ventral and dorsal directions.

Fig. 2

Hierarchical organization of cortical visual areas based on cortical thickness patterns. (A) Pearson’s correlation matrix of mean curvature-corrected cortical thickness across 960 HCP subjects in 25 visual areas. (B) Hierarchical cluster tree of correlation coefficients (fully-connected, clustering threshold = 50th percentile linkage height).

Fig. 3

Group-level variability in mean cortical thickness across 25 visual areas. (A) Density plot of mean curvature-corrected cortical thickness across subjects for each regions of interest. Each point represents the mean cortical thickness for a single subject across both hemispheres, with the centre-of-mass at each visual area indicating the group mean. (B) Gaussian distribution model of group cortical thickness for each visual area. Note the two gradients of increasing cortical thickness along the ventral and dorsal visual areas.

Fig. 4

Hemispheric bias for cortical thickness across 25 visual areas. (A) Mean cortical thickness was calculated for the left and right hemispheres independently for each subject, and the difference (left – right) taken as indicator of hemispheric bias. (B) Mean cortical thickness bias across participants and 95% confidence intervals on the mean show areas that significantly deviate from hemispheric symmetry. Areas V3d, PHC1, IPS2, PS3 and IPS3 showed significant bias for the left hemisphere. Areas PHC2, V3d, V3B, and IPS0 showed significant bias for the right hemisphere. * ​= ​medium effect size, ** ​= ​large effect size.

Fig. 5

Mean surface-corrected cortical thickness for 960 HCP subjects, across 25 visual areas in rank order. The standard deviation of the within-subject cortical thickness estimate, represented by the grey area, is broadly consistent within each visual area. Mean cortical thickness in black, standard deviation in grey.

Fig. 6

Estimates of reliability for mean cortical thickness. In order to estimate within-subject reliability of mean surface-corrected cortical thickness estimates, a leave-p-out resample procedure (1,000 samples, 90% sample size) was conducted, and the 95% confidence interval taken as reliability estimator. One point is displayed per subject. No individual error estimate exceeded 0.1 mm. Dotted lines indicates cortical thickness measurement error as estimated in test-retest studies. Measurement error (1) (Jovicich et al., 2013), measurement error (2) (Han et al., 2006; Madan and Kensinger, 2017).

Appendix A

Quantile-quantile plot of mean curvature-corrected cortical thickness across subjects against the standard normal distribution. Each point represents a single subject across both hemispheres, with the quantile interval represented by a red line. The group-level distribution of mean cortical thickness is consistent with a standard normal distribution across 25 visual areas

Appendix D

Relationship between subject age at time of scanning and mean cortical thickness in 25 visual areas. All cortical areas examined show a negative trend, with cortical thickness decreasing with age at a rate of 0.002 mm (±0.001 SD) per year. Pearson’s correlation estimates (r) calculated across subjects, best linear fit indicated in red

Appendix E

Relationship between self-reported gender and mean surface-corrected cortical thickness across 25 visual areas. One point displayed per subject. No significant effect of gender, or interaction with age, on cortical thickness was detected