Military-to-Civilian Transition Stressors Associated With Couples' Intimate Partner Violence.

Alzheimer’s disease (AD) and Lewy body dementia (LBD) are among the most common neurodegenerative dementias in older adults (; ). Despite distinct underlying pathologies, the two disorders often present with overlapping cognitive and clinical features, particularly in early disease stages. Symptoms such as memory decline, executive dysfunction, and cognitive fluctuations can occur in both conditions, making differential diagnosis challenging (; ; , ). Accurate differentiation between AD and LBD is, therefore, crucial, as the two disorders differ in prognosis, treatment response, and disease mechanisms (). Structural magnetic resonance imaging (MRI) has been widely used to investigate disease-related brain changes in dementia. Traditional morphometric analyses typically focus on regional structural measures, including cortical thickness, cortical gray-matter volume, and subcortical volumetric measures. These approaches have revealed characteristic medial temporal atrophy in AD and relatively preserved hippocampal structures in LBD (; ). However, regional morphometric measures alone may not fully capture distributed structural alterations across the cortex, particularly given the substantial inter-individual variability in cortical morphology (; ; ). Increasingly, connectomics approaches have been proposed to bridge regional structural alterations and large-scale network organization, providing a systems-level perspective on neurodegeneration (; ). Morphometric similarity networks represent one such framework, quantifying the similarity of structural features across cortical regions to characterize coordinated anatomical organization (; ). These networks have been shown to reflect underlying cytoarchitectonic and connectivity patterns and can reveal disease-related alterations in cortical organization (; ; ). Despite their promise, most existing morphometric similarity studies rely on atlas-based cortical parcellations and cross-subject registration (; ; ). Such approaches may obscure individual-specific folding geometry and introduce potential alignment biases when comparing cortical structures across individuals (). Furthermore, atlas-based regions impose a fixed node definition that may not optimally capture fine-scale structural variation. Importantly, morphometric properties of the cortex are themselves strongly shaped by cortical folding geometry. Structural features such as cortical thickness vary systematically along the gyral–sulcal spectrum and are closely related to local curvature and cortical shape (). Cortical folding patterns reflect fundamental neurodevelopmental and biomechanical constraints shaping cortical architecture and are closely linked to underlying connectivity patterns and functional specialization (; ; ). Consequently, cortical folding patterns may provide a natural scaffold for constructing structural similarity networks. Leveraging folding landmarks, therefore, offers an individualized and biologically meaningful basis for defining network nodes and modeling cortical similarity. Recent studies, including our previous work (; ; ), have begun to explore the potential of folding-informed individualized networks for dementia analysis. Our previous work (; ) has demonstrated that individualized structural networks derived from cortical folding landmarks can achieve promising performance in dementia classification using deep learning models. In particular, both cortical similarity networks constructed from morphometric features and fiber connectivity networks derived from diffusion MRI have been used to distinguish AD, LBD, and cognitively normal (CN) individuals using data-driven models. These studies suggest that folding-informed individualized networks capture diagnostically relevant structural patterns. However, predictive modeling alone provides limited insight into the biological mechanisms underlying these classification results. These approaches primarily focus on predictive performance and provide limited insight into the structural mechanisms underlying the observed classification results. In particular, it remains unclear which aspects of network organization contribute to disease differentiation and how network topology relates to the underlying cortical morphology. Understanding these mechanisms is important not only for interpreting predictive models but also for revealing biologically meaningful alterations in cortical organization associated with neurodegenerative diseases.