Quantifying white matter tract diffusion parameters in the presence of increased extra-fiber cellularity and vasogenic edema

Abstract

The effect of extra-fiber structural and pathological components confounding diffusion tensor imaging (DTI) computation was quantitatively investigated using data generated by both Monte-Carlo simulations and tissue phantoms. Increased extent of vasogenic edema, by addition of various amount of gel to fixed normal mouse trigeminal nerves or by increasing non-restricted isotropic diffusion tensor components in Monte-Carlo simulations, significantly decreased fractional anisotropy (FA) and increased radial diffusivity, while less significantly increased axial diffusivity derived by DTI. Increased cellularity, mimicked by graded increase of the restricted isotropic diffusion tensor component in Monte-Carlo simulations, significantly decreased FA and axial diffusivity with limited impact on radial diffusivity derived by DTI. The MC simulation and tissue phantom data were also analyzed by the recently developed diffusion basis spectrum imaging (DBSI) to simultaneously distinguish and quantify the axon/myelin integrity and extra-fiber diffusion components. Results showed that increased cellularity or vasogenic edema did not affect the DBSI-derived fiber FA, axial or radial diffusivity. Importantly, the extent of extra-fiber cellularity and edema estimated by DBSI correlated with experimentally added gel and Monte-Carlo simulations. We also examined the feasibility of applying 25-direction diffusion encoding scheme for DBSI analysis on coherent white matter tracts. Results from both phantom experiments and simulations suggested that the 25-direction diffusion scheme provided comparable DBSI estimation of both fiber diffusion parameters and extra-fiber cellularity/edema extent as those by 99-direction scheme. An in vivo 25-direction DBSI analysis was performed on experimental autoimmune encephalomyelitis (EAE, an animal model of human multiple sclerosis) optic nerve as an example to examine the validity of derived DBSI parameters with post-imaging immunohistochemistry verification. Results support that in vivo DBSI using 25-direction diffusion scheme correctly reflect the underlying axonal injury, demyelination, and inflammation of optic nerves in EAE mice.

Keywords: Diffusion basis spectrum imaging; Diffusion tensor imaging; Experimental autoimmune encephalomyelitis (EAE); Immunohistochemistry (IHC); Inflammation; Magnetic resonance imaging; Monte-Carlo simulation; Multiple tensor model; Restricted diffusion; White matter injury.

Figure 1

(A) A three-dimensional computer-synthesized trigeminal nerve model was constructed for Monte-Carlo simulations. The coherently oriented axonal fiber bundle was composed of tightly packed cylinders of 2-μm diameter (green cylinders), to be described by anisotropic water diffusion components. Cells were represented by spheres of 6-μm diameter (blue spheres) randomly placed surrounding the fiber bundle for simulating restricted isotropic water diffusion in cells. Extra-axonal/extracellular space was represented as non-restricted isotropic diffusion within an imaging voxel. Voxel of interest (50 × 50 × 50 μm³ pink cube) was placed in the center of the simulation space for the random walk of water molecules (light blue outer sphere of 90-μm diameter). (B) Baseline trigeminal nerve model (model # 1) consisted of intra- and extra-axonal water diffusion closely associated with axonal fibers (anisotropic diffusion; green bar), cells (restricted isotropic diffusion; blue bar) and extra-axonal/extracellular space (non-restricted isotropic diffusion; pink bar). Similar to the fixed normal trigeminal nerve phantoms with different amount of gel, by gradually increasing the size of imaging voxel, a set of baseline trigeminal nerve model with varying fractions of non-restricted isotropic diffusion (model # 2-11) was used to assess the impact of edema on DTI and DBSI measurements. By increasing the number of spheres, i.e., restricted isotropic diffusion components (model # 12-21), the impact of increased cellularity on DTI and DBSI was also assessed.

Figure 2

The impact of various extents of vasogenic edema (or tissue loss) on DTI (open symbols) and DBSI (filled symbols) measurements was mimicked experimentally by adding various amount of 2% agar gel to the fixed normal trigeminal nerve (A) or by increasing the extent, i.e., by enlarging the voxel size without changing the cell or axon bundle content, of non-restricted diffusion component in the Monte-Carlo simulation (B). Experimentally (A), increased gel content resulted in overestimating DTI derived λ_∥ and λ_⊥ (comparing with the nerve only phantom) while underestimated FA (open symbols). The DBSI derived diffusion parameters were not affected by varying the content of the gel (filled symbols). On the data generated using Monte-Carlo simulation (B), increasing non-restricted diffusion components also resulted in overestimating DTI derived λ_∥ and λ_⊥ while underestimated the FA (open symbols). The DBSI derived diffusion parameters (filled symbols) were not affected by increasing non-restricted diffusion component.

Figure 3

Due to the difficulty in making reliable experimental phantoms with varying cell contents, Monte-Carlo simulations with varying extents of the restricted diffusion component were performed to mimic the inflammation associated cellularity increase. Increased content of restricted diffusion component significantly underestimated the DTI derived λ_∥ and FA without an impact on λ_⊥. None of the DBSI derived diffusion parameters was affected by the increased restricted diffusion component.

Figure 4

DBSI derived non-restricted (A) and restricted isotropic diffusion fraction (B) using 99- (circle) and 25-direction (triangle) diffusion encoding scheme was compared with the input values used for Monte-Carlo simulations. Data obtained from both diffusion encoding schemes fall on the line of identity (black dashed lines in A and B) suggesting that DBSI analysis can be accurately performed using 25-direction encoding scheme in situations where fiber crossing is not of concern.

Figure 5

Monte-Carlo simulation of 25-direction diffusion weighted data with SNR of non-diffusion weighted image at 10, 20, 30, 40 and 100 was performed on a mild cell infiltration model (model #12). The noise affected the bias and precision of the 25- direction DBSI-derived axial diffusivity (A), radial diffusivity (B), restricted isotropic diffusion tensor fraction (C), and non-restricted isotropic diffusion tensor fraction (D). At SNR=100, DBSI axial diffusivity, radial diffusivity, restricted isotropic diffusion and non-restricted isotropic diffusion fractions were all estimated with high precision indicated by a small interquartile range and absence of outliers. With decreasing SNR, both bias and precision worsened. At SNR = 40, achieved on the current in vivo DBSI examinations of EAE and sham control mice, DBSI-derived λ_ǁ (A), λ_⊥ (B), restricted (C), and non-restricted (D) diffusion fraction were all well-estimated.

Figure 6

Representative in vivo DTI/DBSI parameters maps and post-MRI immunohistochemical staining of the optic nerves from the sham (labeled by capital letters) and EAE-affected mice (labeled by lower case letters) were employed to detect underlying optic nerve pathologies. The decreased λ_ǁ measured by DTI (a) and DBSI (b) matching the SMI-31 staining (g and h) in EAE optic nerve comparing with those seen in the sham (A, B, G, and H) reflected axonal injury. Demyelination was seen as increased λ_⊥ in EAE optic nerves derived by DTI (c) and DBSI (d) and the loss of MBP staining (i and j) comparing with that of the sham (C, D, I, and J). Increased restricted diffusion fraction (e) of EAE matched the pattern of increased DAPI-positive nuclear staining (k and l), comparing with that of the sham optic nerve (E, K, and L). Increased non-restricted diffusion tensor fraction derived by DBSI was also increased (f) in the EAE optic nerve comparing with the sham (E), potentially reflecting the increased vasogenic edema at the onset of optic neuritis. The correspondence between in vivo DBSI and postmortem immunohistochemistry findings supports that the 25-direction diffusion encoding scheme is adequate for assessing optic nerve pathologies in EAE mice. Quantitative analysis of Immunohistochemical staining was performed on the 60× images (H, h, J, j) for SMI-31 and MBP, and on 20× image of DAPI (K, k) Scale bars represent 100 μm (20×), 10 μm (60×).

Figure 7

Consistent with previous findings that in vivo DTI derived λ_ǁ and λ_⊥ correctly reflected optic nerve axonal injury and demyelination (A, and B). During the onset of optic neuritis, inflammation led to the increased cellularity (evidenced by increased DAPI-positive counts and increased DBSI-derived restricted diffusion fraction) and edema (increased non-restricted diffusion fraction detected by DBSI). It is unclear how exactly these two factors counteracted in computation of DTI derived parameters. The current and previous results suggest that DTI detected the underlying axonal injury and demyelination in optic nerves of EAE mice. DBSI derived λ_ǁ, λ_⊥, and the restricted diffusion fraction correlated with SMI-31 (C), MBP (D), and DAPI (E) staining well.