Recent developments in three-dimensional graphical modeling of the musculoskeletal system are directly related to the development of MRI and CT technologies. Solid models of anatomic structures are created from sequential serial sections of two-dimensional (2D) images of the structures. Stacking the slices creates three-dimensional (3D) models from the 2D images. Tools to reconstruct 3D images from the 2D sequential slice data have been developed and widely used in the diagnostic field [14,34]. When two-dimensional picture elements (pixels) are stacked in the third dimension, they become volume elements, referred to as voxels. Voxel-based models require large amounts of storage space and core memory. Also, the rendering speed achievable with this type of model is too slow to allow real-time display and manipulation. "Vector-based" solid models use the voxel image data to generate a descriptive representation of the surfaces of the objects within the model. Both automated edge detection routines and interactive contouring programs are used to map voxel regions into surface models representing anatomical structures. A 3D distribution of nodes provides vertices for a mesh of polygons that define the shape of each structure. The polygon map of the surface requires far less computer memory than a 3D voxel-based structure and can be manipulated with comparative ease [9,10]. These graphical models are being used to measure joint motion (kinematic analysis), joint and muscle loading (kinetic analysis), and the distribution of the contact pressure and ligament tension within a joint.
Both generic and specific 3D models of the musculoskeletal system are used for biomechanical analysis. Three-dimensional graphical models of living subjects are created for preoperative planning and performance assessment. Three-dimensional models of cadaver specimens allow parallel image-based modeling and experimental testing of the musculoskeletal system. Generic models of joints or other anatomic structures are developed from common domain libraries of musculo-skeletal image data. The techniques used to create 3D models from image data are identical for generic and specific models. Creating specific graphic models of anatomical structures requires the time and expense of obtaining CT and/or MRI data. Generic anatomical image data are easier to obtain, but can introduce errors when combined with experimental data from living subjects or cadavers because of anatomical variations.
The most widely used common domain musculoskeletal image data is available through the National Library of Medicine's Visible Human Project. Axial image data sets, including CT and photographic anatomic data, from one female cadaver and one male cadaver are available. Axial CT images of the male were taken at 1-mm intervals throughout the body, with a resolution of 512 pixels by 512 pixels. The axial anatomic images were also taken at 1 mm intervals, with a resolution of 2048 pixels by 1216 pixels. The data set for the female cadaver includes axial anatomic images at 0.33-mm intervals and axial CT data at 1-mm intervals.
Digitization of the sequential slices to outline the anatomical structures of interest is usually performed using a commercial image analysis program, such as ANALYZE (Mayo Foundation, Rochester, MN). Some interactivity generally is required to select the initial seed points and the gray-scale threshold level that the software uses to outline each structure. The imaging software or other software packages can assemble the sequential images, create polygon-based surfaces between the consecutive slices, and smooth the mesh by redistributing the polygons. The solid models can be imported into commercial engineering animation software (VisLab, EAI, Ames, IA) to assign colors and textures to the models and to animate musculoskeletal motion (Fig. 1).
Was this article helpful?