Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Open Meshed Anatomy: Towards a comprehensive finite element hexahedral mesh derived from open atlases

Andy Trung Huynh, Benjamin Zwick, Michael Halle, Adam Wittek, Karol Miller

TL;DR

The paper tackles the challenge of generating high-quality FE brain meshes without manual segmentation by using an atlas-based approach from the Open Anatomy SPL/NAC brain atlas. It details a workflow to convert atlas-labelled voxels into an atlas-based 8-node hexahedral mesh, establish a two-way correspondence for structure-aware property assignment, and integrate with 3D Slicer extensions for convenience. The authors demonstrate the method by solving the EEG forward problem modeled by $- abla (C ( abla u))= f$ with tissue conductivities defined on a property map and a dipole source in the left middle temporal gyrus, solved with MFEM. Results show enhanced visualization and anatomically contextual analysis, with a mesh containing 2,318,585 elements and the capability to query outcomes by anatomical regions, highlighting applicability to rapid, clinically relevant brain simulations.

Abstract

Computational simulations using methods such as the finite element (FE) method rely on high-quality meshes for achieving accurate results. This study introduces a method for creating a high-quality hexahedral mesh using the Open Anatomy Project's brain atlas. Our atlas-based FE hexahedral mesh of the brain mitigates potential inaccuracies and uncertainties due to segmentation - a process that often requires input of an inexperienced analyst. It accomplishes this by leveraging existing segmentation from the atlas. We further extend the mesh's usability by forming a two-way correspondence between the atlas and mesh. This feature facilitates property assignment for computational simulations and enhances result analysis within an anatomical context. We demonstrate the application of the mesh by solving the electroencephalography (EEG) forward problem. Our method simplifies the mesh creation process, reducing time and effort, and provides a more comprehensive and contextually enriched visualisation of simulation outcomes.

Open Meshed Anatomy: Towards a comprehensive finite element hexahedral mesh derived from open atlases

TL;DR

The paper tackles the challenge of generating high-quality FE brain meshes without manual segmentation by using an atlas-based approach from the Open Anatomy SPL/NAC brain atlas. It details a workflow to convert atlas-labelled voxels into an atlas-based 8-node hexahedral mesh, establish a two-way correspondence for structure-aware property assignment, and integrate with 3D Slicer extensions for convenience. The authors demonstrate the method by solving the EEG forward problem modeled by with tissue conductivities defined on a property map and a dipole source in the left middle temporal gyrus, solved with MFEM. Results show enhanced visualization and anatomically contextual analysis, with a mesh containing 2,318,585 elements and the capability to query outcomes by anatomical regions, highlighting applicability to rapid, clinically relevant brain simulations.

Abstract

Computational simulations using methods such as the finite element (FE) method rely on high-quality meshes for achieving accurate results. This study introduces a method for creating a high-quality hexahedral mesh using the Open Anatomy Project's brain atlas. Our atlas-based FE hexahedral mesh of the brain mitigates potential inaccuracies and uncertainties due to segmentation - a process that often requires input of an inexperienced analyst. It accomplishes this by leveraging existing segmentation from the atlas. We further extend the mesh's usability by forming a two-way correspondence between the atlas and mesh. This feature facilitates property assignment for computational simulations and enhances result analysis within an anatomical context. We demonstrate the application of the mesh by solving the electroencephalography (EEG) forward problem. Our method simplifies the mesh creation process, reducing time and effort, and provides a more comprehensive and contextually enriched visualisation of simulation outcomes.
Paper Structure (10 sections, 1 equation, 5 figures, 2 tables)

This paper contains 10 sections, 1 equation, 5 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Objectives and Scope
  3. Methods
  4. Data acquisition and processing
  5. Atlas-based hexahedral FE mesh creation
  6. Solution of the EEG forward problem
  7. Results
  8. Visualisation of Simulation Results
  9. Simulation results
  10. Discussion

Figures (5)

  • Figure 1: Open Anatomy Project’s SPL/NAC Brain Atlas. Top layer: T1-weighted MRI with 1 mm isotropic resolution. Bottom layer: SPL/NAC Brain Atlas label map overlayed with MRI.
  • Figure 2: Brain label maps used for EEG forward problem. Top layer: Open Anatomy label map with additional CSF, skull and scalp labels. Bottom layer: Properties label map for conductivity assignment.
  • Figure 3: Atlas-based hexahedral FE mesh. Left: FE mesh with Open Anatomy labels. Right: FE mesh with Properties labels.
  • Figure 4: Cut-out of atlas-based FE mesh with streamlines of the electric field (in $\mu$V) generated by a current dipole source located inside the left middle temporal gyrus. The mesh elements are coloured according to (left) the anatomical labels and (right) the tissue conductivity compartments.
  • Figure 5: The electric potential (in $\mu$V) queried by the anatomical structure, namely the left middle temporal gyrus (a), left amygdala (b), left parahippocampal gyrus (c) and the left fusiform gyrus (d).