χ-separation insights into whole-brain characterization of age-related patterns of susceptibility in healthy aging

Quantitative Susceptibility Mapping (QSM) enables noninvasive assessment of brain tissue composition, but conventional approaches provide only a composite measure that merges paramagnetic and diamagnetic contributions, limiting biological specificity. Recent advances in χ-separation overcome this limitation by separating χ-paramagnetic (χ-para) and χ-diamagnetic (χ-dia) components within a single voxel. This study aimed to comprehensively characterize age-related trajectories of paramagnetic and diamagnetic susceptibility changes across the adult lifespan, thereby establishing normative reference patterns for interpreting neuropathological alterations. A total of 131 healthy adults (62 males, 69 females; age 21-89 years) underwent multi-echo gradient echo. χ-separation was applied to generate χ-para, χ-dia, and total susceptibility (χ-tot) maps. Median susceptibility was extracted using a customized 69-region parcellation (cortical, subcortical, and white matter regions). Age effects were assessed with linear and non-linear regression analyses. χ-para exhibited positive linear, quadratic, or exponential associations with aging in caudate, putamen, substantia nigra (SN), red nucleus (RN), subthalamic nucleus (STN), thalamic subdivisions, superior frontal areas around the primary motor cortex, parietal, temporal, occipital, limbic, and insular cortices, splenium of corpus callosum (CC), posterior limb of internal capsule (PLIC,) and anterior of corona radiata (CR). |χ-dia| showed negative linear or quadratic declines in genu, body, and splenium of CC, PLIC, anterior and posterior of CR, posterior thalamic radiation, SN, RN, STN, ventral pallidum, pulvinar, and superior frontal regions. By explicitly separating paramagnetic and diamagnetic components, χ-separation provided novel insights into microstructural age-dependent trajectories, offering biologically specific normative references for iron accumulation and myelin decline, with implications for studying neurodegenerative disorders.