INTRODUCTION: Traumatic brain injury (TBI) is a global health concern, yet the biomechanical mechanisms underlying mild TBI remain poorly understood(1). Finite element (FE) head models can estimate brain-tissue strain, but generating models that reflect cohort-specific neuroanatomy and geometry remain challenging. We developed a reproducible MRI-to-model workflow that generates a study-specific FE head model from structural and diffusion MRI. This study-specific reference model complements our existing subject-specific MRI-to-model pipeline, enabling individualized models to be automatically generated from a new participant MRI. METHOD: Seventy-eight male high-school rugby players underwent T1-weighted and multi-shell diffusion MRI (dMRI). After denoising and correction of common artefacts/distortions, fractional anisotropy maps and fibre orientation distributions (FODs) were estimated. A cohort template was generated using ANTs and MRtrix3 from T1, FA, and FOD images and automatically segmented to derive cortical parcellations and major white-matter tracts. Falx and tentorium were transferred from an open dataset (2) and registered to the template. A whole-brain hexahedral mesh was generated using Polycube; an in-house script assigned element-wise anatomical labels; hyper-viscoelastic fibre-reinformed anisotropic models were used to describe brain tissue which were assigned in a region-specific manner using FA values from dMRI based on previous work (3); and fibre orientations were incorporated from the template FOD map. RESULT: The pipeline produced a cohort-derived template FE model that has cortex, cerebellum, brainstem, falx, and tentorium represented as distinct parts. This can serve as a reference mesh, from which subject-specific models can be derived by the free form deformation method (3) and mapping subject-specific fibre orientations from MRI. DISCUSSION: Our work represents reproducible and automated pipeline that can generate both cohort and subject-specific FE head models for head impact and other simulations. This will allow us to combine FE-derived strain metrics with machine learning to link real-world impact exposure to brain structural and tissue dynamic changes. 1. Shim VB et al. IEEE Access. 2020;8:179457–65. 2. Bayly P V et al. Ann Biomed Eng. 2021 Oct;49(10):2677–92. 3. Shim VB et al. Eng Comput. 2022 Oct 12;38(5):3925–37.