Imaging Developmental Trajectories of Laminar Magnetic Susceptibility Throughout Adulthood in Human Brain Cortex

Iron, myelin, and proteins are critical for neural integrity and cognitive function, yet their concentrations and distributions in brain tissue vary significantly across the lifespan. This natural variability necessitates age-specific reference ranges that account for developmental and aging-related changes. Noninvasive imaging techniques, such as Quantitative Susceptibility Mapping (QSM) and R2* imaging, provide complementary insights into these dynamics. Although age-related susceptibility trajectories have been well characterized in subcortical regions, systematic investigations of laminar susceptibility across the entire cortex remain limited. In this study, we analyzed QSM and R2* images from 447 healthy adults (18-80 years) to characterize cortical susceptibility profiles. Rather than averaging values across entire cortical regions, we normalized cortical depth to a percentage scale (1%-100%), where 1% corresponds to the pial surface and 100% to the white matter boundary. Image intensities at varying cortical depths were extracted and averaged to estimate depth-specific iron levels, motivated by known layer-dependent iron distribution in the cortex. Our results support depth-specific spatial patterns of iron in QSM and R2*: QSM peaked in deep cortical layers (80%-99%), with higher frontal and occipital values compared with temporal and parietal regions. Conversely, elevated R2* in superficial temporal and occipital layers suggests region-specific microstructural variations. Both QSM and R2* exhibited quadratic age-related trajectories, increasing until midlife before declining slightly. Unlike in white matter (WM) and deep gray matter (GM), cortical iron and myelin distributions exhibit distinct patterns, rendering conventional susceptibility-iron models for WM/GM unsuitable for cortical regions. To address this limitation, we focused on high-iron cortical layers, where diamagnetic confounds are minimized. By calibrating our in vivo age-QSM data against the ex vivo age-iron reference curve, we developed an optimized linear transformation and used a specific combination of layers across regions. This refined model achieved superior predictive accuracy (R² = 0.43) compared with the standard framework. Our findings underscore the necessity of laminar-specific analysis for establishing accurate cortical susceptibility benchmarks, improving the discrimination of normal aging from neurodegenerative pathology.