Session: Research Posters

Paper Number: 173574

Image-Processing-Informed 3d Finite Element Modeling to Characterize Aging-Induced Neurodegeneration and Atrophy in Brain White Matter 

Aging in the human brain is accompanied by profound microstructural changes that drive cognitive decline and neurodegeneration. As neurons and oligodendrocytes deteriorate over time, the brain exhibits significant atrophy and volume loss, contributing to memory deficits, diminished motor function, and behavioral changes. Anatomical alterations such as cortical thinning, ventricular enlargement, sulcal widening, and demyelination collectively impair neural transmission and are characteristic of neurodegenerative diseases including Alzheimer’s disease, multiple sclerosis, and other dementia-related conditions. Understanding the spatiotemporal progression of these changes is essential for accurate diagnosis, patient-specific intervention, and the development of effective therapeutic strategies.

Magnetic Resonance Imaging (MRI) and Magnetic Resonance Elastography (MRE) have transformed in-vivo assessments of evolving brain morphology by enabling precise, reproducible, and non-invasive measurements of structural degradation. MRI captures volumetric loss and changes in brain micro-architecture, while MRE provides critical insights into mechanical property alterations such as stiffness and shear moduli that evolve with age or injury. These imaging modalities have proven highly sensitive to localized atrophic changes, serving as essential tools in detecting and quantifying neurodegeneration. Nevertheless, imaging data alone cannot fully capture the underlying biomechanics driving tissue degeneration, underscoring the need for advanced computational approaches that can simulate and interpret these complex interactions.

To address this gap, this research integrates detailed MRI and MRE data with in-house developed three-dimensional finite element (FE) micromechanical models designed to simulate the aging brain’s biomechanical response. These models explicitly represent brain white matter (BWM) microstructure by defining representative volume elements (RVEs) that incorporate bionic myelinated axon geometry. Here, axonal tracts are modeled as aligned, fiber-like inclusions embedded within an extracellular glial matrix (ECM), mimicking the natural organization of white matter. This bio-inspired geometric representation enables the FE framework to capture realistic anisotropy, fiber dispersion, and evolving microstructural features observed in aging and neurodegeneration.

The computational approach explores the effects of microstructural geometry, volume fraction (VF), and evolving material properties (shear moduli) on tissue-level stress–strain behavior under uniaxial tensile loading conditions. By incorporating image-derived structural and mechanical property distributions, the models account for multidimensionality, multidirectionality, and inter-individual variability, providing a more realistic representation of brain aging mechanics.

Three distinct FE model types were developed to investigate the influence of key aging-related factors on BWM mechanics. Model Type I isolates the effect of volume fraction changes to assess how reductions in axonal density impact mechanical response. Model Type II focuses on the deterioration of shear moduli in the constituent phases (axon, myelin, ECM) to simulate softening driven by demyelination and cellular degeneration. Model Type III combines both volume fraction reduction and material property evolution to capture higher-order aging-induced neurodegeneration. Stress versus strain analyses across these models revealed that cerebral damage from aging is tightly linked to micro-architectural variations, loading direction, and the evolving mechanical behavior of constituent phases. The results demonstrate significant tissue softening and degeneration patterns consistent with clinical imaging observations, validating the models' ability to replicate observed neurodegenerative changes.

The integration of advanced imaging with finite element modeling offers a powerful, non-invasive approach to interpret aging-induced brain degeneration. These image-informed, data-driven micromechanical FE models can be calibrated and optimized using publicly available brain scan datasets to serve as potential biomarkers for neuroimaging studies and clinical practice. Moreover, they provide a framework for patient-specific predictive modeling, supporting the development of personalized intervention strategies. Overall, this research advances the state of the art in brain biomechanics by providing a comprehensive computational toolset capable of characterizing the complex mechanical interactions driving neurodegeneration and atrophy in aging brain white matter.

 

Presenting Author: Mohit Agarwal Rutgers, The State University of New Jersey

Presenting Author Biography: Mohit Agarwal is a Senior Research Specialist at the Materials Engineering Center (MEC), part of Engineering & Process Science (E&PS) within Core R&D at Dow. He is a computational mechanics engineer with deep expertise in data science-driven finite element modeling (FEM) of composites. His technical strengths span FEA, applied machine learning (ML), optimization, and analytical solid mechanics, enabling the development of hybrid computational models that integrate data and physics for advanced material analysis. Mohit earned his Ph.D. in Mechanical and Aerospace Engineering (MAE) from Rutgers University, where his research focused on combining FEM and ML to model brain white matter and aging biomaterials under traumatic brain injury conditions. He also developed hybrid frameworks that combine FEM and ML to characterize soft material properties, advancing digital thread and digital twins in soft materials research. In addition, he brings over six years of industry experience as a Senior R&D Product Development Engineer and Project Manager in the medical device sector, where he led cross-functional teams and new product introductions (NPDP). He also holds dual master’s degrees in MAE and electrical & computer engineering (Machine Learning specialization) from Rutgers University – New Brunswick, NJ, USA.

Authors:

Mohit Agarwal Rutgers, The State University of New Jersey
John Georgiadis Illinois Institute of Technology
Assimina Pelegri Rutgers, The State University of New Jersey