MRI-based human brain atlases of R1, R2, proton density, and myelin volume fraction using synthetic quantitative imaging at 1.5 T.

Abstract

Quantitative MRI (qMRI) enables objective, reproducible measurement of tissue-specific MR properties, offering improved diagnostic accuracy and deeper insight into disease mechanisms compared to conventional MRI. Parameters such as proton density (PD), relaxation rates (R1, R2), and myelin volume fraction (MVF) provide valuable information on tissue composition, water content, and myelination. Reference qMRI atlases based on these parameters are essential for identifying deviations in individuals and supporting clinical decision-making. While several reference atlases exist, none offer comprehensive, multi-parametric data derived from a single acquisition dataset. In this study, we constructed four synthetic 3D reference brain atlases (isotropic resolution 1.5 mm) of PD, R1, R2, and MVF using the clinically integrated multi-dynamic multi-echo sequence at 1.5 Tesla based on 58 healthy controls (HC). These atlases allow both visual and quantitative comparisons with individual patient data and will be made available to the research community. We validated the atlases using data from four HC and four multiple sclerosis (MS) patients, analyzing both white and gray matter regions. The atlases showed high anatomical accuracy, spatial resolution, and full brain coverage. Significant deviations were observed in MS patients compared to the reference dataset (p < 0.01), while a testing group of HCs closely matched the atlas values. Z-score maps accurately captured lesion patterns and diffuse abnormalities in an individual patient with MS, reflecting early demyelination and tissue damage. These findings demonstrate the clinical potential of our atlases for detecting subtle brain changes.