Lyndia Wu, University of British Columbia

Title: Integrating Real-world Head Impact and Neuroimaging Data into a Computational Pipeline towards Large-scale and Multiscale Modeling of Sports Brain Injury

Abstract

Contemporary biomechanical modeling of traumatic brain injury struggles to bridge the critical gap between global organ-level dynamics and microscale axonal injury. This lecture introduces a novel, integrated computational pipeline designed to overcome this. Leveraging unique, multimodal datasets from ice hockey athletes with diagnosed concussions, our framework creates subject-specific global brain models incorporating detailed white matter anisotropy from imaging. This allows simulation of voxelized brain deformation and white matter fiber strains. A key innovation is the subsequent integration of a convolutional neural network (CNN) axonal injury model. Trained on distinct fiber strain profiles, the CNN efficiently provides whole-brain axonal injury metrics. This enables estimation of injury risk across ~5000 voxels for hundreds to thousands of head impacts an athlete might sustain per season. The pipeline facilitates mechanistic investigation of multiscale brain injury risks, offering significant potential as a diagnostic and cumulative load monitoring tool.