A quantitative multi-parameter mapping protocol standardized for clinical research in multiple sclerosis

Abstract

Quantitative magnetic resonance imaging (qMRI) involves mapping microstructure in standardized units sensitive to histological properties and supplements conventional MRI, which relies on contrast weighted images where intensities have no biophysical meaning. While measuring tissue properties such as myelin, iron or water content is desired in a disease context, qMRI changes may typically reflect mixed influences from aging or pre-clinical degeneration. We used a fast multi-parameter mapping (MPM) protocol for clinical routine at 3T to reconstruct whole-brain quantitative maps of magnetization transfer saturation (MT), proton density (PD), longitudinal (R1), and transverse relaxation rate (R2*) with 1.6 mm isotropic resolution. We report reference MPM values from a healthy population with age and gender distributions typical of multiple sclerosis in whole brain white matter (WM), T2-weighted WM hyperintensities, cortical grey matter and deep grey matter regions and present post-processing optimizations including integration of lesions and normalization of PD maps against cerebrospinal fluid (CSF) for standardized research in multiple sclerosis (MS) and potentially also in related disorders. PD maps were affected by WM abnormalities in MS using WM calibration. The results acknowledge the impact of non-linear age effects on MPM and suggest using CSF calibration for future clinical application in MS.

Keywords: Aging; Multi-parameter mapping; Multiple sclerosis; Neuroimaging; Proton density; Quantitative MRI.

Fig. 1

Graphical representation of the MPM pipeline. Raw PDw, MTw, T1w echoes were corrected for Gibb’s artifact before reconstruction with the hMRI toolbox. Receive field inhomogeneities were corrected using Unified Segmentation. T1-MPRAGE was segmented using Fastsurfer, a deep learning alternative to FreeSurfer, to obtain tissue masks for white matter (WM), cortical grey matter (CGM) and deep grey matter (DGM). White matter T2 hyperintensities were manually segmented from T2-FLAIR. All masks were then spatially registered to the quantitative maps. PD maps were calibrated as 100% in ventricular CSF. Voxels with T1 < 4s were excluded from the eroded lateral ventricles mask.

Fig. 2

(a) PD comparison in whole white matter (WM) and normal-appearing white matter (NAWM) of 27 MS patients between calibration methods using lesion-filled white matter mask (PD with NAWM calibration) and whole white matter mask (PD with whole WM calibration). Significance levels associated to asterisks: p  < 0.05 (*), p  < 0.01 (**), p  <  0.001 (***). (b) Comparison of PD maps calibrated using CSF signal in normal appearing white matter (NAWM) and whole white matter (WM) regions of MS patients. Using CSF calibration, the mean difference between NAWM and WM is 0.1 p.u, which is of the same order of magnitude as in (a). Standard deviations are higher resulting in a higher inter-subject coefficient of variation. (c) Comparison of non-calibrated PD maps in normal appearing white matter (NAWM) and whole white matter (WM) regions of MS patients.

Fig. 3

(a) Histograms of median MPM values distributions across healthy participants in white matter (WM, yellow), white matter lesions (WML, purple), cortical grey matter (CGM, red), deep grey matter (DGM, blue). Dashed lines indicate respective median. For each tissue class except WML, outliers outside the 2–98th percentile were removed. WML only included median values of healthy participants with mean volume higher than 0.2 mL. (b) MPM comparison of white matter lesions in MS patients (MS_LESION) and HC T2w white matter hyperintensities (HC_LESION) against healthy white matter (HC_WM) of healthy controls and normal appearing white matter of MS patients (MS_NAWM, free of lesions). Significance levels associated to asterisks: p  < 0.05 (*), p  <  0.01 (**), p  <  0.001 (***).

Fig. 4

(a) Boxplots and density distribution comparison of median MT, PD, R1 and R2^* values in thalamus, caudate, globus pallidus (Pallidum), putamen, amygdala, hippocampus, nucleus accumbens (Accumbens). (b) Greyscale and RGB-colored slice examples of population averaged quantitative maps showing globus pallidus caudate putamen and thalamus. In particular, the globus pallidus shows higher R2^*, R1, MT and lower PD values.

Fig. 5

Scatterplots and fitted trajectories (blue) for the described ROIs. Green curve shows the linear spline with a cut-off of 55y when it performed better than the other models. Red and green dots represent women and men respectively.