Modern Technology in Multi-Shell Diffusion MRI Reveals Diffuse White Matter Changes in Young Adults With Relapsing-Remitting Multiple Sclerosis

Abstract

Objective: To characterize microstructural white matter changes related to relapsing-remitting multiple sclerosis using advanced diffusion MRI modeling and tractography. The association between imaging data and patient's cognitive performance, fatigue severity and depressive symptoms is also explored.

Methods: In this cross-sectional study, 24 relapsing-remitting multiple sclerosis patients and 11 healthy controls were compared using high angular resolution diffusion imaging (HARDI). The imaging method includes a multi-shell scheme, free water correction to obtain tissue-specific measurements, probabilistic tracking algorithm robust to crossing fibers and white matter lesions, automatic streamlines and bundle dissection and tract-profiling with tractometry. The neuropsychological evaluation included the Symbol Digit Modalities Test, Paced Auditory Serial Addition Test, Modified Fatigue Impact Scale and Beck Depression Inventory-II.

Results: Bundle-wise analysis by tractometry revealed a difference between patients and controls for 11 of the 14 preselected white matter bundles. In patients, free water corrected fractional anisotropy was significantly reduced while radial and mean diffusivities were increased, consistent with diffuse demyelination. The fornix and left inferior fronto-occipital fasciculus exhibited a higher free water fraction. Eight bundles showed an increase in total apparent fiber density and four bundles had a higher number of fiber orientations, suggesting axonal swelling and increased organization complexity, respectively. In the association study, depressive symptoms were associated with diffusion abnormalities in the right superior longitudinal fasciculus.

Conclusion: Tissue-specific diffusion measures showed abnormalities along multiple cerebral white matter bundles in patients with relapsing-remitting multiple sclerosis. The proposed methodology combines free-water imaging, advanced bundle dissection and tractometry, which is a novel approach to investigate cerebral pathology in multiple sclerosis. It opens a new window of use for HARDI-derived measures and free water corrected diffusion measures. Advanced diffusion MRI provides a better insight into cerebral white matter changes in relapsing-remitting multiple sclerosis, namely diffuse demyelination, edema and increased fiber density and complexity.

Keywords: cognition; free-water imaging; high angular resolution imaging; multiple sclerosis; tractometry.

FIGURE 1

MRI processing pipeline. (A) Inputs diffusion-weighted images are processed by the TractoFlow pipeline. (B) Local modeling of the diffusion bi-tensor with tissue and free-water compartments. The fODF are estimated using constrained spherical deconvolution. (C) Diffusion MRI-derived measures and free-water fraction are computed. (D) Raw T1-weighted and T2-FLAIR images. (E) Manual and automatic lesions segmentations are performed in order to allow tracking through lesions. (F) Probabilistic tractography is computed in the lesion-corrected white matter mask. (G) Extraction of the preselected white matter bundles by RecoBundles. (H) Tractometry of each bundle using the previously computed diffusion measures and free-water fraction.

FIGURE 2

Tracking through white matter lesions. Manual and automatic segmentations are fused to produce the “lesion mask” (in yellow), covering all white matter lesions visible on the T2-FLAIR and T1-weighted (A) MRI images. Probabilistic tractography based on fODF peaks (C) is then executed and tracking is allowed through lesions, as shown on the FA map (B). Without this correction, no track would be generated inside the yellow mask.

FIGURE 3

Fornix bundle-specific segmentation. Adapted from Rheault et al., 2018. Axial view (A) and sagittal view (B) of the regions of interest used for the manual segmentation of the fornix. Mamillary bodies in purple (inclusion, 1), body of the fornix in yellow (inclusion, 2), both thalami in red (exclusion, 3) and both hippocampi in green (inclusion, 4). In thumbnail, the general morphology of the segmented fornix is presented.

FIGURE 4

Selected white matter bundles of interest. White matter bundles obtained from the tractogram of a RRMS patient (female, 23 years old, disease duration of 5 years, EDSS 1.5, treated with Natalizumab). Extraction with RecoBundles, which uses bundle models as shape priors for detecting similar tracks and bundles in tractograms. Tracks segmented from the participant’s data were visually assessed to ensure quality of the bundle extraction. (A) Anterior corpus callosum (rostrum, genu); (B) Posterior corpus callosum (isthmus, splenium, tapetum); (C) Fornix; (D) Cingulum; (E) Inferior fronto-occipital fasciculus; (F) Inferior longitudinal fasciculus; (G) Superior longitudinal fasciculus; (H) Uncinate fasciculus.