Neurite density index is sensitive to age related differences in the developing brain

Highlights

• We show evidence supporting the use of NODDI instead of FA in the developing brain.

• We provide a direct comparison between Neurite density index (NDI) and DTI metrics.

• NDI is better at classifying subject age group than FA.

• NDI explains more variance in white matter with age than FA.

• Over development NDI is more sensitive to age related white matter differences.

Abstract

Purpose

White matter development during childhood and adolescence is characterised by increasing white matter coherence and organisation. Commonly used scalar metrics, such as fractional anisotropy (FA), are sensitive to multiple mechanisms of white matter change and therefore unable to distinguish between mechanisms that change during development. We investigate the relationship between age and neurite density index (NDI) from neurite orientation dispersion and density imaging (NODDI), and the age-classification accuracy of NDI compared with FA, in a developmental cohort.

Method

Diffusion-weighted imaging data from 72 children and adolescents between the ages of 4–19 was collected (M=10.42, SD=3.99, 36 male). We compared NODDI metrics against conventional DTI metrics (fractional anisotropy [FA], mean diffusivity [MD], axial diffusivity [AD] and radial diffusivity [RD]) in terms of their relationship to age. An ROC analysis was also performed to assess the ability of each metric to classify older and younger participants.

Results

NDI exhibited a stronger relationship with age (median R²=.60) compared with MD (median R²=.39), FA (median R²=.27), AD (median R²=.14), and RD (median R²=.35) in a high proportion of white matter tracts. When participants were divided into an older and younger group, NDI achieved the best classification (median area under the curve [AUC]=.89), followed by MD (median AUC=.81), FA (median AUC=.80), RD (median AUC=.81), and AD (median AUC=.64).

Conclusion

Our results demonstrate the sensitivity of NDI to age-related differences in white matter microstructural organisation over development. Importantly, NDI is more sensitive to such developmental changes compared to commonly used DTI metrics. This knowledge provides justification for implementing NODDI metrics in developmental studies.

Introduction

Brain maturation across childhood and adolescence is one of the most dynamic and important periods in the development of the brain (Paus et al., 1999, Sowell et al., 2003). Understanding typical development of the brain in children and adolescents, when, where and why maturational changes occur is important for a better understanding of the localization, connectivity and maturation of brain function, cognition, and behavior. This understanding also establishes a baseline from which to reveal when and how neurodevelopmental processes go awry.

A number of neuroimaging studies have sought to characterise the cortical grey matter changes over development (Giedd et al., 1999, Wierenga et al., 2014). White matter volume has consistently been shown to increase throughout adolescence and into adulthood but little is known about the underlying microstructural processes causing this volume change or the relationship with function. Magnetic Resonance Imaging (MRI) techniques such as diffusion-weighted imaging (DWI) provide the ability to indirectly examine the microstructural components of white matter, with age-related changes in diffusion metrics thought to relate to neurobiological processes including myelination and axonal organization (Beaulieu, 2002, Paus, 2010).

The tensor model is most commonly used to derive white matter microstructure metrics including fractional anisotropy (FA), mean, axial and radial diffusivity (MD, AD and RD) (Basser and Pierpaoli, 1996, Mori and Zhang, 2006). FA, the most frequently used measure, has been shown to increase over childhood and adolescence (Lebel et al., 2008, Lebel et al., 2012, Simmonds et al., 2014). This increase of FA with age is typically attributed to increased myelination as the white matter matures, however FA is a relatively non-specific metric, and can also be influenced by white matter organisation, axonal density, as well as both intra- and extra-cellular mechanisms (Beaulieu, 2009). The recently developed neurite orientation dispersion density imaging (NODDI) is a multi-compartment model of white matter microstructure, and models the biophysical properties of white matter (Zhang et al., 2012). It offers orientation dispersion index (ODI) and neurite density index (NDI) as alternative metrics to FA. These two indices aim to better quantify, and disentangle, neurite morphology in the brain. ODI models the intra-neurite space (between axons) to characterise angular variation of neurites as well as cell membranes, somas and glial cells that influence the extra-neurite (extracellular) space. NDI models intra-neurite space and characterises density of neurites by restricted diffusion (Sepehrband et al., 2015). Being a more sophisticated model of underlying neurobiology, these measures might reveal more about the developing brain.

Given the relatively recent development of this technique, few studies have investigated NODDI metrics over development. Chang et al. (2015) revealed that across the lifespan, from childhood to late adulthood, there is a strong relationship between NDI and chronological age, compared with FA. Other studies have investigated the relationship between NDI and pre-term birth (Kelly et al., 2016), early development (Jelescu et al., 2015), and ageing in adulthood (Kodiweera et al., 2016, Merluzzi et al., 2016), but have not directly compared NODDI and DTI metrics.

Here we investigate the relationship between DTI and NODDI metrics over white matter development in 72 children and adolescents between 4 and 19 years of age. The main aims of this study were to: (1) model age-related differences in diffusion metrics in development; and (2) compare the sensitivity and specificity of NDI and ODI against DTI metrics over age using an ROC analysis.