Automated Locus Coeruleus segmentation and age-related signal changes across Neuromelanin-sensitive MRI and T1/T2-weighted images

Abstract

The locus coeruleus (LC) plays a key role in stress response, attention, and cognitive function, with its integrity closely linked to neuromelanin accumulation. Neuromelanin-sensitive MRI (NM-MRI) has enabled in vivo imaging of the LC, but automated segmentation remains challenging due to its small size, low contrast, and anatomical variability. This study aimed to evaluate the accuracy of automatic LC segmentation using multi-atlas nonlinear warping applied to high-resolution T1/T2-weighted images and to compare LC intensity characteristics across NM-MRI, T1w, and T2w images. Additionally, we examined developmental changes in LC signal from childhood through adulthood. We analyzed MRI data from 134 participants (ages 8[~]64) with NM-MRI, T1w, and T2w scans, alongside 652 participants (ages 6[~]22) from the Human Connectome Project in Development (HCP-D). LC intensity measures were compared across imaging modalities and analyzed for age-related trends. Automatic LC segmentation achieved high spatial overlap with manual tracings (up to 95% after one voxel 2D-dilation). Statistically estimated high (or low)-intensity values (mean {+/-} 2{sigma}) from automated LC masks provided a more stable and noise-resilient alternative to local peak intensity measures. Age-related LC signal changes are most clearly observed in NM-MRI and T2w imaging, with significant developmental shifts emerging during adolescence. These findings support the conventional T2w imaging as an indirect marker of neuromelanin accumulation and highlight the possibility of age-specific analyses in LC imaging studies.