Research Papers

Development and Validation of Patient-Specific Finite Element Models of the Hemipelvis Generated From a Sparse CT Data Set

[+] Author and Article Information
Vickie B. Shim1

Bioengineering Institute, University of Auckland, 1010 New Zealandv.shim@auckland.ac.nz

Rocco P. Pitto

Department of Orthopaedic Surgery, and Bioengineering Institute, University of Auckland, 1142 New Zealand

Robert M. Streicher

 Stryker SA., Thalwil 8800, Switzerland

Peter J. Hunter, Iain A. Anderson

Bioengineering Institute, University of Auckland, 1010 New Zealand

An interactive computer program developed by the Bioengineering Institute for Continuum Mechanics, Image analysis, Signal processing and System identification. Freely available for academic use.



Corresponding author.

J Biomech Eng 130(5), 051010 (Aug 14, 2008) (11 pages) doi:10.1115/1.2960368 History: Received March 26, 2007; Revised May 06, 2008; Published August 14, 2008

To produce a patient-specific finite element (FE) model of a bone such as the pelvis, a complete computer tomographic (CT) or magnetic resonance imaging (MRI) geometric data set is desirable. However, most patient data are limited to a specific region of interest such as the acetabulum. We have overcome this problem by providing a hybrid method that is capable of generating accurate FE models from sparse patient data sets. In this paper, we have validated our technique with mechanical experiments. Three cadaveric embalmed pelves were strain gauged and used in mechanical experiments. FE models were generated from the CT scans of the pelves. Material properties for cancellous bone were obtained from the CT scans and assigned to the FE mesh using a spatially varying field embedded inside the mesh while other materials used in the model were obtained from the literature. Although our FE meshes have large elements, the spatially varying field allowed them to have location dependent inhomogeneous material properties. For each pelvis, five different FE meshes with a varying number of patient CT slices (8–12) were generated to determine how many patient CT slices are needed for good accuracy. All five mesh types showed good agreement between the model and experimental strains. Meshes generated with incomplete data sets showed very similar stress distributions to those obtained from the FE mesh generated with complete data sets. Our modeling approach provides an important step in advancing the application of FE models from the research environment to the clinical setting.

Copyright © 2008 by American Society of Mechanical Engineers
Your Session has timed out. Please sign back in to continue.



Grahic Jump Location
Figure 1

(a) Locations of the five strain gauges used in the experimental setup; (b) harvested pelvis with soft tissue removed

Grahic Jump Location
Figure 2

Schematic of our experimental setup

Grahic Jump Location
Figure 3

(a) Mesh generated, (b) all of the Gauss points inside the mesh, and (c) close up of a mesh section showing detailed material property assignment to each Gauss point

Grahic Jump Location
Figure 4

Algorithm for obtaining cortical regions. A CT image from the acetabular region is shown on the left along with its medial axis (also called the skeleton of the image). Rays drawn are represented with red lines. Density profiles obtained from the rays are shown on the right. The red regions are the regions identified to be cortical bones.

Grahic Jump Location
Figure 5

Results of applying the cortical thickness measuring algorithm to simple shapes and actual CT scans. For simple geometries, circular and square shapes were used and randomly generated noise was included to the images to mimic actual CT scans. For the actual CT scans, we used Visible Human CT scans.

Grahic Jump Location
Figure 6

A flowchart that explains how to assign material properties to FE meshes from CT scans. The overall FE simulation procedure is explained in Fig. 7.

Grahic Jump Location
Figure 7

A flowchart describing the procedures involved in FE simulation. It describes the steps in FE simulation from the CT scans. Step 4 is further explained in Fig. 6.

Grahic Jump Location
Figure 8

Scatter plots from one of the pelves. The horizontal axis is for experimental strains and the vertical axis for the FE strains. All meshes show a good agreement between FE simulated strain values and experimental strain values.

Grahic Jump Location
Figure 9

Von Mises stress distribution plots from one of the pelves tested. Meshes generated with a hybrid method using a sparse patient data set produced very similar results as the mesh generated with a full patient data set.

Grahic Jump Location
Figure 10

Simulation of bending of composite beam. (a) Schematic and loading condition; (b) mesh showing grid points inside.

Grahic Jump Location
Figure 11

Results of four-point bending simulation with a composite beam. (a) shows that analytic solution was reached as DOFs increased; (b) shows the stress distribution at a cross section of the beam. The finite element stress distribution was similar to the analytic solution.

Grahic Jump Location
Figure 12

The four meshes used for convergence testing. The model stabilized when the DOFs reached 7056, hence the mesh density at that level was chosen for simulation.

Grahic Jump Location
Figure 13

A convergence curve for the FE simulation setup. When DOFs reach around 7000, the curve stabilized and reached an asymptote.



Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In