Cortical maturation in the preterm period revealed using a multi-component diffusion-weighted MR model

Target Audience: Researchers or clinicians with an interest in diffusion imaging as applied to a preterm neonatal cohort. Introduction/Purpose: The human brain undergoes radical transformation late in gestation and cortical folding occurs largely during this period. In the cortex, radial glial cells act as scaffolding for neural migration, which has been imaged using the diffusion tensor model. The cortical fractional anisotropy (FA) drops markedly during the preterm period as dendrites are elaborated and water diffuses more isotropically. However, FA is a summary measure and the same parameter value can represent a range of microstructure. Neurite Orientation, Dispersion and Density Imaging (NODDI) 2 uses a multi-compartment model, with multi-shell acquisition, to fit parameters relating to geometric properties and neuronal packing. Thus it disentangles contributions to the FA or other DT parameters and we gain greater insight into the microstructure. Very preterm infants (VPT = <32 weeks completed gestation) are vulnerable to crucial maturational processes being disrupted. By imaging cortical microstructure with the NODDI model, we aim to identify normal and abnormal patterns of growth so as to facilitate earlier and more targeted intervention. Methods: 13 VPT infants (25.6 ± 0.9 weeks gestational age (GA)) with normal cerebral ultrasound imaging were scanned soon after birth (32.0 ± 1.9 weeks GA) and at term equivalent age (41.7 ± 3.7 weeks GA). We acquired DWI in a 3T Philips MRI scanner with 6 volumes at b = 0smm, 16 at b = 750 smm, 32 at b = 2000 smm (48 diffusion directions in total); resolution = 1.75x1.75x2.00 mm, TR = 9s and TE = 60ms, with total duration 11m43s. We removed motion-corrupted volumes and eddy-current corrected the remaining data, rotating the b-vectors and modulating by the expansion/contraction of the transformation (Jones). We used the ‘AdaPT’ algorithm to segment the cortical regions of the brain. We fitted the NODDI model in the cortex and evaluated the three volume fractions: vic (intra-cellular), vec(extra-cellular) and viso(isotropic, representing CSF), each of which has its own signal equation. The intra-cellular signal depends on the orientation dispersion index (ODI) – a summary measure of how dispersed the local structure is.