Correlations between inter-subject variability in tissue properties of human V1, V2, and V3

2106 

Abstract Submission 

Maiko Uesaki^1,2, Toshikazu Miyata^1,3, Noah Benson⁴, Jonathan Winawer⁵, Hiromasa Takemura^1,3,6

¹Center for Information and Neural Networks (CiNet), Advanced ICT Research Institute, NICT, Suita, Japan, ²Graduate School of Frontier Biosciences, Osaka University, Suita, Japan, ³National Institute for Physiological Sciences, Okazaki, Japan, ⁴University of Washington, Seattle, WA, United States, ⁵New York University, New York City, NY, United States, ⁶The Graduate Institute of Advanced Studies, SOKENDAI, Hayama, Japan

Maiko Uesaki  
Center for Information and Neural Networks (CiNet), Advanced ICT Research Institute, NICT|Graduate School of Frontier Biosciences, Osaka University
Suita, Japan|Suita, Japan

Toshikazu Miyata  
Center for Information and Neural Networks (CiNet), Advanced ICT Research Institute, NICT|National Institute for Physiological Sciences
Suita, Japan|Okazaki, Japan

Noah Benson  
University of Washington
Seattle, WA, United States

Jonathan Winawer  
New York University
New York City, NY, United States

Hiromasa Takemura  
Center for Information and Neural Networks (CiNet), Advanced ICT Research Institute, NICT|National Institute for Physiological Sciences|The Graduate Institute of Advanced Studies, SOKENDAI
Suita, Japan|Okazaki, Japan|Hayama, Japan

Over the past several decades, neuroanatomy and neuroimaging research has revealed large individual differences in the size of the human visual cortex (Stensaas et al. 1974; Andrews et al. 1997; Dougherty et al. 2003). Further studies have documented covariance amongst the size of visual areas, and between the size of visual areas and properties of other visual structures (Benson et al. 2022; Miyata et al. 2022). We wondered whether there was also substantial variability in the grey matter tissue microstructure of the visual cortex across individuals, and how such measures covary throughout the multiple cortical maps.. Recent advances in structural neuroimaging provide opportunities for characterising tissue properties of cortical areas using MRI. The ratio of T1- to T2-weighted signal intensity (T1w/T2w) has become a widely used semi-quantitative measure of tissue microstructure of the brain (Glasser & van Essen, 2011; Berman et al. 2022). Here, we analysed the Human Connectome Project 7T retinotopy dataset (Benson et al. 2018), to evaluate individual differences and covariance of T1w/T2w amongst early visual areas V1, V2, and V3.

Analyses were performed on the 3T structural and 7T functional MRI data acquired from 160 subjects who participated in the retinotopic mapping experiments (Benson et al. 2018) as part of the Human Connectome Project (van Essen et al. 2012). Borders of V1, V2, and V3 (regions of interest; ROIs) were manually drawn by four researchers based on polar angle reversals in the retinotopic map (Benson et al. 2022). We identified the mid-grey surface of ROIs for each subject using neuropythy (https://github.com/noahbenson/neuropythy), and averaged T1w/T2w across voxels belonging to each ROI. We first calculated inter-subject correlations between T1w/T2w of ROIs. Subsequently, we computed how much of the individual variability in T1w/T2w of each ROI could be predicted by that of the entire cortex using linear regression. We then used the residuals to evaluate whether, and the extent to which, T1w/T2w covaried amongst ROIs once the cortex-wide variability was partialled out.

Figure 1 depicts correlations between T1w/T2w of ROIs across subjects for each hemisphere. Strong correlations were found between V1 and V2 (left: r = 0.92; right: r = 0.92), V2 and V3 (left: r = 0.93; right: r = 0.95), and V1 and V3 (left: r = 0.89; right: r = 0.91). T1w/T2w tended to be higher in V1, compared with V2 and V3, consistent with known anatomical differences in grey matter microstructure between V1 and V2/V3. We note that T1w/T2w of corresponding ROIs were also highly correlated between hemispheres (left vs right hemispheres for V1: r = 0.90; V2: r = 0.90; V3: r = 0.84). The high correlations were not simply due to some subjects having overall high or low T1w/T2w values in all of cortex. Figure 2 shows correlations between the residuals of ROIs after T1w/T2w of the whole-brain grey matter was regressed out. Strong correlations remained between V1 and V2 (left: r = 0.87; right: r = 0.86), V2 and V3 (left: r = 0.89; right: r = 0.90), and V1 and V3 (left: r = 0.82; right: r = 0.83).

We examined whether tissue properties, reflected in T1w/T2w, covaried between areas V1, V2, and V3 across 160 individuals. Results showed strong correlations between the border areas V1 and V2, as well as V2 and V3, and also between V1 and V3 though to a lesser extent, in both hemispheres, which persisted even when adjusted for cortex-wide T1w/T2w. T1w/T2w of corresponding areas in the two hemispheres were also highly correlated. Our findings generally align with the finding that the surface areas of V1, V2, and V3 are highly correlated within and across hemispheres (Benson et al. 2022), and suggest that structural covariance between early visual areas is not limited to their surface area, but also found in grey matter tissue microstructure.

Cortical Anatomy and Brain Mapping ¹

Perception: Visual ²

STRUCTURAL MRI

Vision