Hybrid Diffusion Imaging to Detect Acute White Matter Injury after Mild TBI
Sourajit Mitra Mustafi1, Chandana Kodiweera2, Laura A. Flashman3, Thomas W. McAllister4, and Yu-Chien Wu1

1Department of Radiology and Imaging Sciences, Indiana University School of Medicine, Indianapolis, IN, United States, 2Department of Psychological and Brain Sciences, Dartmouth College, Dartmouth, NH, United States, 3Department of Psychiatry, Dartmouth Hitchcock Medical Center and New Hampshire Hospital, Lebanon, NH, United States, 4Department of Psychiatry, Indian University School of medicine, Indianapolis, IN, United States


In the present study we used multi-shell Hybrid Diffusion Imaging (HYDI) to study white matter changes in the acute stage of mild traumatic brain injury (mTBI). Nineteen mTBI patients and 23 trauma-controlled subjects were recruited and studied within 1 month of injury. Non-parametric diffusion analysis, q-space imaging as well as parametric analyses including conventional DTI and novel neurite orientation dispersion and density imaging (NODDI) were used to analyze the HYDI data. Only intra-axonal volume fraction of the NODDI model showed significant and diffuse decrease in white matter of the mTBI patients.


Mild traumatic brain injury (mTBI) is an enormous public health problem that can be associated with prolonged neurobehavioral sequelae. Currently mTBI is diagnosed based on clinical assessment, which often includes self-report. As yet there is no objective imaging biomarker of sufficient sensitivity and specificity to be considered a valid and reliable indicator of mTBI. Self-report measures of mTBI may be unreliable1, heightening interest in neuroimaging biomarkers that could be used to more accurately diagnose mTBI and assist in monitoring recovery and treatment. Diffusion MRI (dMRI) measures water diffusion behaviors in biological systems, and is sensitive to microstructural changes in white matter (WM). Recent developments in dMRI enable direct links of diffusion measurements to WM microstructures. We used multi-shell Hybrid Diffusion Imaging (HYDI)2 for non-parametric diffusion analysis, q-space imaging2, and parametric analyses including conventional DTI and novel neurite orientation dispersion and density imaging (NODDI)3. NODDI hypothesizes WM microstructures in three compartments: extracellular, intracellular and cerebrospinal fluid compartment. The evaluated NODDI parameters may explain cellular level changes in mTBI.


Participants: Nineteen mTBI patients and 23 trauma-controlled subjects were recruited from the Emergency Department. The demographics of the subjects are listed in Table 1. The MRI studies were performed in the acute stage, within 1 month of injury. MRI: Participants received T1W SPGR and HYDI in a Philips 3T Achieve TX scanner with 8-channel head coil and SENSE parallel imaging. The diffusion-weighting (DW) pulse sequence was a SS-SE-EPI sequence. TR was 3.59 sec. Other MR parameters were: TE = 114.24 ms, δ/Δ=46/58.4 ms, voxel size=2*2 mm2, 40 slices with slice thickness =3 mm, parallel-imaging SENSE factor=2. The total scan-time was about 24 min. Similar to reference 2, the diffusion encoding scheme consisted of one b0 and 5 b-value shells (250, 1000, 2250, 4000, and 6250 s/mm2) with total 142 DW directions (6, 21, 24, 30, and 61 in each shell). Image processing: The whole dataset was processed with a non-parametric approach, q-space imaging, and yielded P0, a tissue restriction index2,4,5. The first and second shell was processed for DTI, which produces diffusion measures including fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD). A parametric approach using the NODDI model used the whole HYDI dataset. NODDI produces intracellular (i.e., intra-axonal in WM cases) volume fraction (Vic), extracellular volume fraction (Vec), volume fraction (Viso) of isotropic fast diffusion (e.g., CSF), and fiber orientation dispersion index (ODI)3. The NODDI fitting was performed using AMICO-NODDI package6. WM ROI: Forty-eight WM ROIs were defined in the standard MNI space by intersecting subjects’ mean WM skeleton with WM atlas of Johns Hopkins University (JHU) ICBM-DTI-817 (Figure 1). All diffusion measures were non-linearly transformed to standard MNI space with FSL FNIRT8. Statistics: Linear model analysis was used to test the significance of diffusion metrics between mTBI and trauma-controlled groups with sex and gender as covariates (model 3 in Table 2). Multiple comparisons across 48 ROIs were adjusted using false discovery rate (FDR) with significant level at q-value < 0.05.

Results and Discussions

Maps of DTI, q-space and NODDI diffusion metrics of an mTBI subject are shown in Figure 2. Among various diffusion metrics, only the NODDI derived intracellular volume fraction (Vic) was sensitive to mTBI with significant decreases in 60% of WM ROIs (Table 2). The mTBI subjects had an approximately 4% decrease in Vic. Given the NODDI microstructural assumption of rigid-stick axons (fixed intra-axonal axial diffusivity and near-zero intra-axonal radial diffusivity), Vic represents parenchymal axonal density (1-Viso). In addition, significant decreases of Vic were accompanied by significant increases of Vec (i.e., Vec = 1 - Vic). Therefore, the results suggest that in the acute stage of mTBI, the density of stick-like axons decreases and the inter-axonal space increases. Furthermore, the organization of axons described by DTI fractional anisotropy (FA) and NODDI orientation dispersion index (ODI) remains unchanged. The overall tissue restriction described by P0 and DTI axial and radial diffusivities did not show significant difference between groups. The prevalence of Vic-significant ROIs described a diffuse change in WM in the acute phase of mTBI (Table 2). The affected WM tracts concentrated on pyramidal tracts and its cortical projections (bilateral corona radiatae). Most of the cerebella related tracts and hippocampal tracts are spared.


HYDI and its diffusion metrics provide insights about microstructural changes of WM in the acute stage of mTBI and may prove useful as a marker of injury.


The work is supported by NIH R21 NS075791 and IUPUI-ITDP grants.


1. Ruff RM, Iverson GL, Barth JT, Bush SS, Broshek DK, Policy NAN, et al. Recommendations for diagnosing a mild traumatic brain injury: a National Academy of Neuropsychology education paper. Archives of clinical neuropsychology : the official journal of the National Academy of Neuropsychologists. 2009;24(1):3-10.

2. Wu YC, Alexander AL. Hybrid diffusion imaging. Neuroimage. 2007;36(3):617-29. PMCID: 2428345.

3. Zhang H, Schneider T, Wheeler-Kingshott CA, Alexander DC. NODDI: practical in vivo neurite orientation dispersion and density imaging of the human brain. Neuroimage. 2012;61(4):1000-16.

4. Assaf Y, Mayk A, Cohen Y. Displacement imaging of spinal cord using q-space diffusion-weighted MRI. Magn Reson Med. 2000;44(5):713-22.

5. Ozarslan E, Koay CG, Shepherd TM, Komlosh ME, Irfanoglu MO, Pierpaoli C, et al. Mean apparent propagator (MAP) MRI: a novel diffusion imaging method for mapping tissue microstructure. Neuroimage. 2013;78:16-32. PMCID: 4059870.

6. Daducci A, Canales-Rodriguez EJ, Zhang H, Dyrby TB, Alexander DC, Thiran JP. Accelerated Microstructure Imaging via Convex Optimization (AMICO) from diffusion MRI data. Neuroimage. 2015;105:32-44.

7. Oishi K, Zilles K, Amunts K, Faria A, Jiang H, Li X, et al. Human brain white matter atlas: identification and assignment of common anatomical structures in superficial white matter. Neuroimage. 2008;43(3):447-57. PMCID: 2586008.

8. Smith SM, Jenkinson M, Johansen-Berg H, Rueckert D, Nichols TE, Mackay CE, et al. Tract-based spatial statistics: voxelwise analysis of multi-subject diffusion data. Neuroimage. 2006;31(4):1487-505.


Figure 1: WM ROIs defined as intersection of mean FA skeleton with JHU white matter atlas in standard MNI space. A. The mean FA skeleton overlaid on the mean FA map. B. JHU white matter atlas. C. The WM ROIs. The acronyms for ROIs are listed in Table 2.

Figure 2: DTI, P0 and NODDI maps for an mTBI subject is shown. The gray scales of AD, RD and MD are 0 to 3, 2 and 2*10-3 mm2/s, respectively. The FA, P0, ODI, Vic and Viso has been scaled from 0 to 1.

Table 1. Subject demographics.

Table 2: Statistic results of Vic across 48 white matter ROIs. Green denotes decrease in mTBI group. Sig denotes significant with FDR q-value < 0.05. NS denotes non-significant.

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)