More insights into early brain development through statistical analyses of eigen-structural elements of diffusion tensor imaging using multivariate adaptive regression splines

Yasheng Chen, Hongtu Zhu, Hongyu An, Diane Armao, Dinggang Shen, John H. Gilmore, Weili Lin

    Research output: Contribution to journalArticlepeer-review

    4 Citations (Scopus)

    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 data-driven 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, full-term 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 eigen-structural elements of DTI.

    Original languageEnglish
    Pages (from-to)551-569
    Number of pages19
    JournalBrain Structure and Function
    Volume219
    Issue number2
    DOIs
    Publication statusPublished - 2014 Mar

    Bibliographical note

    Funding Information:
    This study was supported in part by NSF Grant BCS-08-26844 and NIH Grants RR025747-01, P01CA142538-01, MH086633, and AG033387, NIH Grants 1R01EB006733, R01EB008374, and 1R01EB009634, NIH Grants R01MH070890 and R01HD053000, and NIH Grant R01NS055754.

    Keywords

    • DTI longitudinal analysis
    • DTI regression analysis
    • Early brain development
    • GEE
    • Multivariate adaptive regression splines

    ASJC Scopus subject areas

    • Anatomy
    • General Neuroscience
    • Histology

    Fingerprint

    Dive into the research topics of 'More insights into early brain development through statistical analyses of eigen-structural elements of diffusion tensor imaging using multivariate adaptive regression splines'. Together they form a unique fingerprint.

    Cite this