Image-based meshing

Image-based meshing is opening up exciting new possibilities for the application of computational continuum mechanics (numerical methods such as Computational fluid dynamics (CFD) and Finite Element analysis to problems in Biomechanics, Soil mechanics, Material characterization, and Nondestructive testing. Meshing techniques that can rapidly generate robust, high quality meshes from complex 3D image data, as can be obtained from Magnetic resonance imaging (MRI), Computed tomography (CT) or Microtomography for example, are increasingly in demand. Different methods of generating the required volume discretizations directly and robustly from the image data have been developed, however there are a range of issues related to image processing and mesh generation which still need to be addressed.

Mesh generation from 3D imaging data

Although a wide range of mesh generation techniques are currently available these, on the whole, have not been developed with meshing from segmented 3D imaging data in mind. Meshing from 3D imaging data presents a number of challenges but also unique opportunities for presenting more realistic and accurate geometrical description of the computational domain. The majority of approaches adopted have involved generating a surface model (either in a discretized or continuous format) from the scan data, which is then exported to a commercial mesher – so-called ‘CAD-based approach’. This process is often time consuming, not very robust and virtually intractable for the complex topologies typical of image data. A more direct way is the ‘Image-based approach’ as it combines the geometric detection and mesh creation stages in one process which offers a more robust and accurate result than meshing from surface data. Image-based mesh generation raises a number of issues which are different from CAD-based model generation:

CAD-based approach

CAD-based approaches use the scan data to define the surface of the domain and then create elements within this defined boundary. Although reasonably robust algorithms are now available, these techniques do not easily allow for more than one domain to be meshed, as multiple surfaces are often non-conforming with gaps or overlaps at interfaces where one or more structures meet.

Image-based approach

This approach combines the geometric detection and mesh creation stages in one process. The technique has been pioneered by Simpleware, and generates 3D hexahedral or tetrahedral elements throughout the volume of the domain, thus creating the mesh directly with conforming multipart surfaces. In the case of modeling complex topologies with possibly hundreds of disconnected domains (e.g. inclusions in a matrix), approaching the problem via a CAD-based approach is virtually intractable. By contrast treating the problem using an Image-based approach is remarkably straightforward, robust, accurate and efficient.

Generating a model

The steps involved in the generation of models based on 3D imaging data are:

Scan and image processing

An extensive range of Image processing tools can be used to generate highly accurate models based on data from 3D imaging modalities, e.g. MRI, CT, MicroCT (XMT), and Ultrasound. Features of particular interest include: • Segmentation tools (e.g. Thesholding, Floodfill, Level set methods, etc.) • Filters and Smoothing tools (e.g. Volume and topology preserving smoothing).

Volume and surface mesh generation

The Image-based meshing technique allows the straightforward generation of meshes out of segmented 3D data. Features of particular interest include: • Multi-part meshing (mesh any number of structures simultaneously) • Mapping functions to apply material properties based on signal strength (e.g. Young's modulus to Hounsfield scale) • Smoothing of meshes (e.g. topological preservation of data to ensure preservation of connectivity, and volume neutral smoothing to prevent shrinkage of convex hulls) • Export to FEA and CFD codes for analysis (e.g. nodes, elements, material properties, contact surfaces)

Typical use

• Biomechanics and design of Medical and dental implants • Food sciences • Forensic science • Materials science (composites and foams) • Nondestructive testing (NDT) • Paleontology and Functional morphology • Reverse engineering • Soil science and Petrology

Publications

Young et al, 2008. An efficient approach to converting 3D image data into highly accurate computational models. Philosophical Transactions of the Royal Society A, 366, 3155-3173.

External links

Simpleware Ltd. (www.simpleware.com