Pooled analysis of multiple sclerosis findings on multisite 7 Tesla MRI: Protocol and initial observations

Abstract

Although 7 T MRI research has contributed much to our understanding of multiple sclerosis (MS) pathology, most prior data has come from small, single-center studies with varying methods. In order to truly know if such findings have widespread applicability, multicenter methods and studies are needed. To address this, members of the North American Imaging in MS (NAIMS) Cooperative worked together to create a multicenter collaborative study of 7 T MRI in MS. In this manuscript, we describe the methods we have developed for the purpose of pooling together a large, retrospective dataset of 7 T MRIs acquired in multiple MS studies at five institutions. To date, this group has contributed five-hundred and twenty-eight 7 T MRI scans from 350 individuals with MS to a common data repository, with plans to continue to increase this sample size in the coming years. We have developed unified methods for image processing for data harmonization and lesion identification/segmentation. We report here our initial observations on intersite differences in acquisition, which includes site/device differences in brain coverage and image quality. We also report on the development of our methods and training of image evaluators, which resulted in median Dice Similarity Coefficients for trained raters' annotation of cortical and deep gray matter lesions, paramagnetic rim lesions, and meningeal enhancement between 0.73 and 0.82 compared to final consensus masks. We expect this publication to act as a resource for other investigators aiming to combine multicenter 7 T MRI datasets for the study of MS, in addition to providing a methodological reference for all future analysis projects to stem from the development of this dataset.

Keywords: 7 T; multicenter; multiple sclerosis.

FIGURE 1

Data processing scheme. MP2RAGE, magnetization prepared 2 rapid acquisition gradient echo; Gd, post‐contrast; INV1, 1st inversion; INV2, 2nd inversion; T1‐w, T1‐weighted; T1‐w UNI, uniform T1‐weighted; denoised T1‐w, T1w UNI × INV2_N4 (N4, bias‐field corrected); FLAIR, fluid attenuated inversion recovery; GRE, gradient recalled echo; mag, magnitude; QSM, quantitative susceptibility mapping; LTS‐dFLAIR, LTS‐Gd‐FLAIR − FLAIR (LTS, least trimmed squares); prcnt‐dFLAIR, 100 × (LTS − dFLAIR)/FLAIR.

FIGURE 2

Cortical lesion masking example. Shown is a cortical lesion (yellow arrow) identified on T1‐w (a) and confirmed on FLAIR (b) and ME‐GRE (c). The lesion was then masked (d) as its borders appeared on T1‐w.

FIGURE 3

PRL identification. The yellow arrow indicates a WML identified on T1‐w (a), which is confirmed to meet criteria as a PRL on PHS (b) and QSM (c). The lesion is masked as it appears on T1‐w (d), with the propagation of this mask to PHS (e) and QSM (f).

FIGURE 4

MCE identification. Pre‐contrast FLAIR (a), post‐contrast FLAIR (b), and a FLAIR subtraction map (c) were reviewed for the presence of hyperintensities that met criteria for MCE. The yellow arrow indicates the location of a focus of subarachnoid spread/fill LME, which was then masked on the subtraction map (d).

FIGURE 5

T1‐w comparison. Shown are the denoised T1‐w images from the MP2RAGE acquisition from UMB (a), NINDS (b), UPenn (c), MNI (d), and BWH (e). Differences in gray‐white contrast, brain coverage, and field inhomogeneity can be seen between sites.

FIGURE 6

FLAIR comparison. Shown are comparisons of the 3D FLAIR acquisition from UMB (a), NINDS (b), UPenn (c), and BWH (d). FLAIR not available from MNI. Note differences in tissue contrast, field inhomogeneity, and brain coverage between sites.

FIGURE 7

ME‐GRE comparison. Shown are examples of ME‐GRE from UMB (a), UPenn (b), MNI (c), and BWH (d). ME‐GRE from NINDS not available. Note differences in tissue contrast, field inhomogeneity, and brain coverage between sites.