Abstract

Understanding how mechanical forces translate into progressive neural pathology remains a central challenge in mitigating, diagnosing, and treating traumatic brain injury (TBI), a major public health concern that affects millions annually and lacks reliable prognostic biomarkers. To address this, we combined virtual brain twins constructed using finite element methods with diffusion imaging to investigate spatiotemporal tissue alterations in two porcine TBI models: non-impact rotational TBI and controlled cortical impact (CCI). In the rotational model, subacute abnormalities were observed in the brainstem and subcortical regions, with quantitative anisotropy showing a moderate correlation with peak strain. In contrast, CCI produced delayed, but widespread, diffusion abnormalities, particularly in mean diffusivity, that were associated with cortical deformation over time. Axonal injury metrics derived from tractography-based structural connectivity showed strain-related trends in the rotational model but not in CCI, suggesting that diffuse axonal injury from inertial loading may be more directly influenced by mechanical factors, whereas axonal damage in focal impact may be more influenced by secondary, non-mechanical processes, such as neuroinflammation. These findings support a mechanistic association between simulated brain deformation and evolving imaging biomarkers, indicating that different TBI mechanisms (diffuse versus focal) are associated with distinct spatiotemporal diffusion signatures. Our study provides a translational framework for interpreting neuroimaging through the lens of biomechanics and highlights the potential of virtual brain twins to inform mechanism-specific diagnosis, monitoring, and therapeutic development in TBI.