Date Published: April 29, 2019
Publisher: Public Library of Science
Author(s): Patrick McCunn, Kyle M. Gilbert, Peter Zeman, Alex X. Li, Michael J. Strong, Ali R. Khan, Robert Bartha, Aaron Jon Newman.
Neurite Orientation Dispersion and Density Imaging (NODDI) is a diffusion MRI (dMRI) technique used to characterize tissue microstructure by compartmental modelling of neural water fractions. Intra-neurite, extra-neurite, and cerebral spinal fluid volume fractions are measured. The purpose of this study was to determine the reproducibility of NODDI in the rat brain at 9.4 Tesla.
Eight data sets were successfully acquired on adult male Sprague Dawley rats. Each rat was scanned twice on a 9.4T Agilent MRI with a 7 ± 1 day separation between scans. A multi-shell diffusion protocol was implemented consisting of 108 total directions varied over two shells (b-values of 1000 s/mm2 and 2000 s/mm2). Three techniques were used to analyze the NODDI scalar maps: mean region of interest (ROI) analysis, whole brain voxel-wise analysis, and targeted ROI analyses (voxel-wise within a given ROI). The coefficient of variation (CV) was used to assess the reproducibility of NODDI and provide insight into necessary sample sizes and minimum detectable effect size.
CV maps for orientation dispersion index (ODI) and neurite density index (NDI) showed high reproducibility both between and within subjects. Furthermore, it was found that small biological changes (<5%) may be detected with feasible sample sizes (n < 6–10). In contrast, isotropic volume fraction (IsoVF) was found to have low reproducibility, requiring very large sample sizes (n > 50) for biological changes to be detected.
The ODI and NDI measured by NODDI in the rat brain at 9.4T are highly reproducible and may be sensitive to subtle changes in tissue microstructure.
Diffusion weighted magnetic resonance imaging (dMRI) is a powerful magnetic resonance modality that provides a wealth of information regarding tissue microstructure, from which structural connectivity and pathological changes within the brain can be inferred [1,2]. As different microstructures predictably retard diffusion, the apparent diffusion of molecules combined with the angle of an applied diffusion gradient provides an indirect measure of neuroanatomy . The most commonly used dMRI technique is diffusion tensor imaging (DTI). For DTI, a series of pulsed-gradient, spin–echoes are used to produce a 3×3 symmetric matrix modelling Gaussian diffusion (3). Most commonly, DTI characterizes the overall water diffusion within a given voxel by measuring mean diffusivity (MD) and the degree of directionality of the principle component of this diffusion, through fractional anisotropy (FA). This technique has been utilized for many years and has provided valuable insights into the effects of disease, as well as neurological and physiological processes [4–8].
The minimum accepted average whole-brain SNR was 25 at b = 0 for each of the included data sets. Two data sets were removed from the analysis due to low SNR causing significant reconstruction bias. Therefore, data were successfully acquired and analyzed from eight subjects (age 102 ± 13 days at time of initial scan, weight 323 ± 37 g) at two separate time points. For each subject the re-scan time point was between six and eight days after the original scan. The time of day was not standardized for the scans. Fig 1 shows representative cross sections of raw diffusion data (b = 0) from a single subject, as well as scalar maps of ODI, NDI, and IsoVF.
This study examined the reproducibility of the three most commonly derived NODDI metrics (ODI, NDI, and IsoVF) in the rodent brain at 9.4 Tesla. ODI and NDI were reproducible, showing low coefficients of variation in both the between and within subject conditions. CVs were lower within subjects compared to between subjects, indicating less variability on a within subject scan-rescan basis, as expected. These trends were observed in the mean ROI, whole brain voxel-wise, and targeted voxel-wise analyses.