The Association Between White Matter Tracts and Executive Function in Six Year Old Children Using Robust Scale-Invariant Canonical Correlation Analysis

• Activity Number: 576 - Brain Connectivity Studies
• Type: Contributed
• Date/Time: Wednesday, July 31, 2019 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Imaging
• Author(s): Benjamin Langworthy* and Jason Fine and John Gilmore and Rebecca Stephens
• Companies: University of North Carolina - Chapel Hill and University of North Carolina - Chapel Hill and University of North Carolina - Chapel Hill and University of North Carolina - Chapel Hill
• Keywords: Canonical Correlation Analysis; Diffusion Tensor Imaging; Executive Function; Robust Statistics; White Matter Tractography; Kendall’s Tau

Abstract

We use a scale-invariant robust estimator for canonical correlation analysis (CCA) to estimate the canonical directions and correlations between 16 white matter tracts measured using diffusion tensor imaging and five executive function (EF) tests among six year old children. Our estimator uses a transformation of Kendall’s correlation to estimate the scatter matrix for data from elliptical or transelliptical distributions. This method is more robust than standard CCA to outliers and heavy tailed distributions and is invariant to monotonic transformations of the data. We show that the estimates of the canonical directions and correlations based on this method are consistent and asymptotically normal when the true canonical correlation is greater than zero. We also show through simulations that inverting a bootstrapped confidence interval gives a valid test of the null that a true canonical correlation is zero. Using our method for CCA estimates we show a significant association between lateralization of white matter tracts and EF scores. Lower radial diffusivity in tracts on the right side of the brain is correlated with higher EF scores among children in our sample.