Uncovering Neuropsychiatric Susceptibility Through Reliable and Harmonized Sulcal Morphometry : A Framework for Reproducible Neuroimaging Analysis

Magnetic resonance imaging (MRI) has transformed the study of human brain structure by enabling quantitative analysis of cortical anatomy. However, conventional morphometric measures, such as cortical thickness and volumetry, often lack sensitivity to subtle structural alterations and are vulnerable to scanner- and site-related variability in large, multisite studies. Sulcal morphometry, which characterizes fine- grained aspects of cortical folding geometry, offers a biologically grounded and potentially more sensitive approach for investigating neuroanatomical susceptibility, yet its reliability and harmonizability at scale remain insufficiently established. This dissertation develops and validates a reproducible and harmonized framework for sulcal morphometry, integrating cortical surface reconstruction (FreeSurfer), sulcal feature extraction (BrainVisa), and multisite harmonization (ComBat). Test–retest reliability and cross- version reproducibility were systematically evaluated across multiple datasets (HCP, KKI, OASIS-1), demonstrating that key metrics— particularly sulcal depth and hull junction—are robust across software versions. Scanner- and site-related variability were further quantified and effectively mitigated using empirical Bayes harmonization, enabling a unified morphometric feature space for multisite analysis. This framework was applied to the LIMBIC-CENC cohort of U.S. veterans and service members with and without mild traumatic brain injury (mTBI). In 1,706 post-9/11 veterans and service members, a history of combat-related TBI was associated with wider sulci in seven right- lateralized frontal and anterior-temporal regions after adjustment for age, sex, and intracranial volume (Cohen's d = 0.14–0.22). The associations were confined to the sulcal-width measure, were robust to psychiatric and injury-history confounders, and are consistent with peri-sulcal tissue loss at biomechanically vulnerable cortical folds. Together, this work establishes sulcal morphometry as a reliable and harmonizable neuroimaging phenotype, underscores the importance of methodological standardization for large-scale reproducible analysis, and demonstrates that harmonized sulcal width can reveal subtle, anatomically specific structural signatures of combat-related brain injury that conventional group-level morphometry had left undetected.