Submillimeter postmortem and in vivo diffusion and susceptibility magnetic resonance imaging to characterize cortical...

The human cerebral cortex consists of anatomically and functionally distinct regions defined by heterogeneous cytoarchitectonic, myelin-related microstructural organization. Quantitative magnetic resonance imaging (MRI) provides a noninvasive framework to resolve subtle regional differences across the cortical mantle. Among different MRI contrasts, diffusion and susceptibility MRI of the brain provide complementary sensitivity to micro-, meso-, and macro-structural organization. However, it remains unclear which fine-scale cortical patterns, typically observed within postmortem scans, are preserved in vivo at matched submillimeter resolution, given limited acquisition time and physiological noise. In this study, we performed a comparative, atlas-based quantification of high-resolution human brain MRI images acquired postmortem and in vivo at approximately 0.5 mm isotropic resolution, focusing on diffusion fractional anisotropy (FA) and tissue magnetic susceptibility (χ) derived from quantitative susceptibility mapping (QSM). For postmortem imaging, diffusion MRI was performed on the left hemisphere from a 71 y/o male using a 3T Connectome 2.0 system, while multi-orientation susceptibility MRI was performed at 7T. In vivo datasets were acquired with diffusion MRI using the signal-to-noise ratio-efficient, distortion-free Romer-EPTI acquisition from a healthy volunteer, 27 y/o female, on the same 3T Connectome, and susceptibility MRI on a 7T system at matched submillimeter resolutions. Using OpenMAP-T1 with the JHU-MNI (Eve) Level 4 parcellation, we performed atlas-based analyses on FA and χ, with additional cortical surface mapping to characterize micro- and meso-structural heterogeneities within the cortical mantle. We then assessed the correspondence between postmortem and in vivo measured metrics. Atlas-based cortical surface mapping revealed consistent anatomical variations in both quantitative MRI metrics, with consistently elevated FA and χ values in the precentral and posterior cortical regions relative to surrounding association cortices, while divergent patterns in the cingulate cortex (higher FA with lower χ values) were also observed. Region-wise comparisons between postmortem and in vivo metrics in the left hemispheric cortical regions demonstrated moderate correspondence for FA and χ values (Pearson's r = 0.52 [p = 0.02] and 0.53 [p = 0.01], respectively), despite existing biological and technical variations. Together, these findings provide some initial insight into how quantitative diffusion and susceptibility MRI contrast patterns may translate from high-fidelity postmortem imaging to in vivo acquisitions. Combined submillimeter postmortem and in vivo diffusion and susceptibility MRI in the present study can provide a comprehensive anatomical reference for cortical microstructure mapping and support future translations between postmortem and in vivo neuroimaging studies.