Your Brain Mesh
Automated subject-specific brain finite element mesh creation pipeline
Overview
A reproducible research software project for automated subject-specific finite element brain meshes creation from MRI brain scans. Developed with the funding support from Imperial College London's ReCoDE programme, the project translates a complex image-to-mesh workflow into a clearer notebook-based pipeline with supporting documentation.
GitHub repository
The open-source repository contains the notebook-based workflow, helper scripts, example structure, and setup notes for creating subject-specific finite element brain meshes.
View GitHub repository →Documentation site
The GitHub Pages documentation explains the pipeline stages, expected folder structure, input and output files, and visualisation steps for checking generated meshes.
View documentation →From MRI to mesh
The pipeline starts from subject-level neuroimaging data and anatomical segmentations. You can create your own brain FE mesh if you have your brain scan!
Anatomically detailed model
The resulting models contain approximately one million hexahedral elements, allowing cortical folding and internal anatomical compartments to be represented in three dimensions.
Hybrid MRI segmentation
Combines FreeSurfer parcellation with FSL FAST and BET to generate a complete whole-head segmentation from T1-weighted MRI. This hybrid strategy preserves detailed brain anatomy while filling compartments such as CSF, skull and skin.
Quality-controlled mesh refinement
Direct voxel-to-element conversion preserves anatomical detail but creates stair-step boundaries. The pipeline applies Laplacian smoothing with quality-controlled node locking, followed by contact repair to preserve a continuous CSF layer between brain and skull before simulation.
Reproducible modelling starts before the simulation.