More insights into early brain development through

Authors: Yasheng Chen Hongtu Zhu Hongyu An Diane Armao Dinggang Shen John H Gilmore Weili Lin

Publish Date: 2013/03/01

Volume: 219, Issue: 2, Pages: 551-569

PDF Link

Abstract

The aim of this study was to characterize the maturational changes of the three eigenvalues λ1 ≥ λ2 ≥ λ3 of diffusion tensor imaging DTI during early postnatal life for more insights into early brain development In order to overcome the limitations of using presumed growth trajectories for regression analysis we employed Multivariate Adaptive Regression Splines MARS to derive datadriven growth trajectories for the three eigenvalues We further employed Generalized Estimating Equations GEE to carry out statistical inferences on the growth trajectories obtained with MARS With a total of 71 longitudinal datasets acquired from 29 healthy fullterm pediatric subjects we found that the growth velocities of the three eigenvalues were highly correlated but significantly different from each other This paradox suggested the existence of mechanisms coordinating the maturations of the three eigenvalues even though different physiological origins may be responsible for their temporal evolutions Furthermore our results revealed the limitations of using the average of λ2 and λ3 as the radial diffusivity in interpreting DTI findings during early brain development because these two eigenvalues had significantly different growth velocities even in central white matter In addition based upon the three eigenvalues we have documented the growth trajectory differences between central and peripheral white matter between anterior and posterior limbs of internal capsule and between inferior and superior longitudinal fasciculus Taken together we have demonstrated that more insights into early brain maturation can be gained through analyzing eigenstructural elements of DTI

Keywords: