A VIRTUAL MODELING APPROACH FOR GENERATION, DESIGN ANALYSIS, AND FABRICATION OF 3D HUMAN SKELETAL MODELS

Charles S. Sullivan III and Denis J. DiAngelo

Department of Biomedical Engineering, The University of Tennessee Health Science

Center, Memphis, TN, USA E-mail: ddiangelo@utmem.edu Web: http://memphis.mecca.org/bme/bttlweb/

INTRODUCTION Computer modeling of human anatomy forms the basis of several research approaches in biomedical engineering, including generic model manufacture, manufacture of implant devices, real-time animation, finite element modeling (FEM), and image guided surgery. The expense of and incompatibility between different computer aided design (CAD) and manufacturing (CAM) programs, added with the complex shape and morphology of human anatomy, have limited the creation of accurate virtual models of the human skeletal anatomy and related implant devices (Duerstock, 2003). The Virtual Models can be used in various biomedical engineering applications: manufacture of human skeletal models and joint implants, FEM analysis of human skeletal models and joint implants, simulation and animation of biomechanical test results, and the mechanical design and FEM analysis of implant devices. The practice of combining computer tomography (CT) scans to form a 3D image of human anatomy is not new (Yoo, 2001). Using these images to produce dimensionally accurate models of complex human anatomies such as the spine, skull, jaw, teeth, hands, and feet is currently the subject of several research projects (Enciso, 2003). In this project, we generated a software \ hardware pipeline for the creation of accurate models of human skeletal anatomy.

METHODS A method to generate a model of human skeletal anatomy suitable for use on a desktop PC was developed using the following process: • Identify and evaluate sources for

generating input data for the models. • Identify and evaluate software and

hardware for creating virtual models from input data.

• Generate the virtual 3D solid models. • Demonstrate potential applications. RESULTS AND DISCUSSION Several software packages were evaluated for the model generation process. CT slices in the DICOM (.dcm) format were most commonly used for model generation, although laser surface scanning was also used. Laser scans provide an accurate point cloud of an object, but can only be used to model surface information. Laser scanning of skeletal anatomy cannot be performed on a living patient, which limits some uses for immediate patient benefit. An advantage of CT scans is that they describe both the internal and external surfaces of an object in question and they can be taken from a living patient. Medical image-processing software (SliceOmatic 4.2, Tomovision) was used to segment regions of interest (ROI) from CT scanner data. The process was performed using manual and automated tools that manipulated gray scale image thresholds for tissue separation (Figure 1).

Figure 1: Example of ROI segmentation. Output from the CT segmentation was converted into stereo lithography file formats (.stl) as point clouds. The point cloud data were imported into a volumetric surface modeling program (Geomagic Studio 5, Raindrop). Digital smoothing algorithms, in combination with manual editing techniques within this software, allowed for excellent surface reproduction from point cloud data, as shown in Figure 2.

Figure 2: 3D model from point cloud data. Geomagic Studio also provided for the conversion of the surface models into several file formats (.stl, iges, .stp, .dwg, .3do, etc.) that are compatible with many standard parametric modeling programs as well as popular FEA\FEM and CAD\CAM software. In addition, Geomagic has the ability to add datum points and axes to the models that can be used in “down stream” editing applications. Parametric modeling programs (Pro-Engineer, PTC Corp. and Unigraphics NX, EDS Inc.) were used to edit, assign material properties, and generate tool paths for CNC machining to the models, as shown in Figures 3 and 4 respectively.

Figure 3: Left, Example of the addition of bounding boxes, axes, and textures. Right, 3D solid model of human skeletal anatomy. Figure 4: Examples of FEA and CNC tool path generation of a tooth modeled from CT data using appropriate software. SUMMARY A cost effective method for generating 3D models of human skeletal anatomy using commercially available software suitable for use on a desktop PC was developed. Optimal input data file formats (.dcm) were identified, as well as optimal output formats for surface generation (.stl) and 3D surface modeling (.iges). Software packages were identified that accomplished ROI segmentation, point cloud generation, surface smoothing\editing, and incorporation into parametric modeling programs. REFERENCES Enciso R. et al. (2003) Orthod Craniofacial Res, 6 (Suppl. 1): 66-71. Duerstock B. et al., (2003). Journal of Microscopy, 210: 138-148. Seung-Schik Yoo et al. (2001). Journal of Neuroscience Methods, 112: 75-82.