Bridging Neuroimaging and Pathology in Dementia: A Multi‐Cohort Investigation of MRI‐Derived Brain Volumes

BACKGROUND: The ability to precisely characterize neurodegenerative diseases at early stages remains a challenge. Understanding the relationship between magnetic resonance Imaging (MRI)‐derived brain volumes and neuropathological outcomes is key to advancing early and accurate diagnosis of neurodegenerative diseases. While MRI‐based volumetrics are widely used to assess structural changes in vivo, their link to neuropathological findings remains underexplored. This study examines associations between region‐specific brain atrophy and neuropathological markers, including Alzheimer's Disease Neuropathologic Change (ADNC), CERAD score, Braak staging, and Thal amyloid phase, leveraging volumetric data across multiple aging cohorts.

METHOD: T1‐weighted MRI scans from participants in longitudinal dementia cohorts (ROS, MAP, MARS, and NACC) were pre‐processed, and brain volumes of 120 cerebral regions of interest (ROIs) were calculated using Multi‐Atlas Segmentation Utilizing Ensembles (MUSE). Associations between ROI volumes and neuropathological metrics were analysed using linear regression, adjusting for age at scan, difference between age at death and age at scan, sex, intracranial volume, and site/scanner. Neuropathological metrics, including ADNC, CERAD scores, Thal phase, Braak staging, and vascular metrics like atherosclerosis were analysed as continuous variables grouped categorically (e.g., Thal phase recoded into three categories: 0, 1, and 2). Statistical significance was determined using Benjamini‐Hochberg method ( p <0.05).

RESULT: Among 755 participants (mean age at scan: 80.57±10.23 years; mean age at death: 85.49±10.45 years; 44% male), ADNC was associated with atrophy in temporal and parietal regions, notably middle and inferior temporal gyri, angular gyrus, and amygdala. CERAD scores correlated with atrophy in middle temporal, para‐hippocampal, and fusiform gyrus, Thal phase with atrophy in middle temporal, angular gyrus, and precuneus and Braak staging showed atrophy in entorhinal cortex, middle temporal gyrus, and para‐hippocampal regions. Atherosclerosis was associated with atrophy in prefrontal cortex, and parieto‐occipital regions.

CONCLUSION: These findings highlight the potential of MRI‐derived volumetrics as non‐invasive biomarkers of underlying Alzheimer's pathology. Our findings demonstrate notable atrophy in medial temporal, lateral temporoparietal, and midline parietal regions, supporting recognized AD patterns of neurodegeneration, consistent with pivotal studies by Habes et al.,(2016), Dickerson et al.,(2009). Associations between amyloid, tau burdens and atrophy in temporal, parietal, and limbic regions underscore their diagnostic value in neurodegeneration.