Introduction

By enabling increased spatial resolution and contrast, ultra-high field (UHF) MRI offers new opportunities to investigate multiple sclerosis (MS) in the brain¹. Because of several technical challenges, only a handful of studies have started exploring the benefits of UHF for cervical spinal cord (SC) imaging on healthy volunteers, using anatomical^2-4 and quantitative imaging^4-5. Clinical applications of 7T SC MRI are still in their early stages², with studies reporting enhanced MS lesion detection using anatomical imaging in the cervical⁶ and thoracic⁷ SC, when compared to optimized protocols at 3T. High-resolution multiparametric quantitative MRI (mp-qMRI) is only emerging at UHF^8-9, yet could provide new insights on the pathological SC in MS¹⁰. In this study, a high-resolution mp-qMRI protocol, including T₁ mapping and Diffusion Tensor Imaging (DTI), was employed at 7T to characterize the cervical SC in a cohort of MS patients at the early stage of the disease and to further investigate pathological features and structural damage.

Methods

Experiments were performed on a 7T whole-body research system (Siemens Healthineers) equipped with an eight-channel transceiver coil array (Rapid Biomedical). Seven MS patients at the early stage of the disease (6women/1man, age: 30±7yo, EDSS scores: 1.5±1 [0-3], median disease durations <3yo) and seven healthy controls (5w/2m, age: 24±5yo) were enrolled. The protocol included high-resolution anatomical imaging using T₂*-weighted MEDIC sequence, T₁ mapping relying on two 3D transverse MP2RAGE acquisitions⁸ and DTI with a reduced-FOV diffusion-weighted single-shot spin-echo EPI sequence⁹. Acquisition parameters are provided in Figure_1. All patients were also examined at 3T using conventional MRI. MS lesions were first identified at 3T and then at 7T based on T₂*-weighted images and T₁ maps (consensus between a neurologist and a neuroradiologist). Mp-qMRI post-processing^4,8-9 was done using MATLAB (The Mathworks), PropSeg¹¹, the Spinal Cord Toolbox¹² and FSL¹³ (FMRIB). 3D T₁ maps were resliced and registered to DTI maps at each cervical level. C1-to-C7 T₁ maps were manually segmented in four regions of interest (ROI): normal appearing (NA) GM and WM, GM and WM lesions. GM/WM cross sectional areas (CSA) were measured on T₁ maps. Statistical analyses were performed using all-pairs Wilcoxon (CSA) and multiple comparison Steel-Dwass (mp-qMRI) non-parametric tests on JMP9 (SAS). Evolutions of T₁ with distance from meninges were also investigated.

Results

Figure_2 presents the main findings in terms of lesion detection and delineation from the 7T/3T qualitative radiological investigation. Lesions were mostly found in lateral (65%) and posterior WM (26%), comparable to the litterature¹⁴. Examples of DTI maps, resliced T₁ maps and manually segmented ROI at every cervical level are shown on Figure_3. Overall, 1/3 of the DTI data had to be discarded because of artifacts, mostly at lower levels. Mp-qMRI and statistical analyses results, reported in Figure_4, showed increased T₁, decreased λ_//, λ_┴ and MD in MS lesions located within the WM. Interestingly, only increased T₁ was reported for GM lesions. As for NA tissue, no statistically significant differences relative to healthy controls were found. Figure_5 reports mean CSA measured on the two cohorts, with lower CSA found for patients; and mean T₁ values plotted versus the distance from the outer cerebrospinal fluid or the central canal, with different T₁ evolutions for lateral and posterior NAWM tissue in MS patients.

Discussion

The images obtained at 7T had very high in-plane resolution, enabling improved MS lesions delineation and exquisite anatomical details visualization, including central canal, anterior fissure and posterior septum, which were in majority larger in patients compared to healthy controls (see Figure 2). Figure_5 also suggests these enlargements as a possible cause for increased T₁ values in proximate lateral and posterior NAWM. Although not specific, the T₁/DTI results indicate MS-related microstructural impairments (axonal loss, demyelination) along with related WM atrophy, as previously reported by the litterature¹⁵. It is worth noting that DTI differences were less sensitive than T₁ changes¹⁶ for MS lesions due to several confounding factors, especially for GM. In terms of spatial coverage, the C1-to-C7 10-mm slices used for analyses sampled 75% of the cervical SC on average. Full exploitation of the 3D MP2RAGE volumes will be subject to future work to obtain an exhaustive characterization of the whole cervical SC. Additional patients and matched healthy controls will be included to confirm these results. Future work is also aimed towards improving the robustness of 7T mp-qMRI (sources of artifacts included movements, partial volume effect and poor B₀ shimming) and fully demonstrating its advantages as compared to mp-qMRI at clinical lower fields.

Conclusion

Overall, these preliminary results showed the great potentials of high-resolution mp-qMRI for MS investigations at UHF. By using T₁ mapping especially, specific anatomical features and microstructural characterization were enhanced (atrophy with enlargement of the median fissure, dilatation of central canal and septum).

Acknowledgements

This work was supported by the following funding sources: ARSEP Foundation “Fondation pour l’Aide à la Recherche sur la Sclérose En Plaques”, Investissements d’Avenir 7T-AMI-ANR-11-EQPX-0001, and CNRS (Centre National de la Recherche Scientifique).

References

¹Inglese et al., Expert Review of Neurotherapeutics 2018. ²Barry et al., NeuroImage 2017. ³Sigmund et al., NMR in Biomed 2012. ⁴Massire et al., NeuroImage 2016. ⁵Kogan et al., NeuroImage 2013. ⁶Dula et al., Multiple Sclerosis 2016. ⁷Lefeuvre et al., ISMRM 2016. ⁸Massire et al., ISMRM 2018. ⁹Massire et al., Magn Reson Med 2018. ¹⁰Dula et al., NMR Biomed 2016, ¹¹De Leener et al., NeuroImage 2016. ¹²De Leener et al., NeuroImage 2014. ¹³Jenkinson et al., NeuroImage 2012. ¹⁴Kearney et al., Neurology 2016. ¹⁵Kearney et al., Mult Scler 2014. ¹⁶Mottershead et al., J Neurol 2003.