We are developing modeling and physical simulation tools
for human joints reconstructed from the CT, MRI, and photograph
imaging data sets of the visible human. The objectives are
non-invasive diagnostic tools, systematic preoperative planning,
custom prosthesis design, and quantitative models of strain
injuries. The research challenges are to compute the physics
directly from the data with minimal user intervention and to
compute at interactive speed despite the great size of the data
sets. We address these challenges by hierarchical modeling,
distributed computing, and specialized physical simulations.
The modeling tools of our system construct geometric spline and
finite element models of bones from the CT data and
tendon/ligaments from the MRI and photographic data. The
geometric models are extended to capture material properties,
such as mass distribution, elasticity, and friction coefficients.
The elasticity and friction coefficients are approximated using
qualitative mathematical functions of the CT and MRI intensities.
The simulation module computes the joint physics (motions, forces,
deformations) from the bone models, material properties, and
muscle forces. The visualization module displays the raw data and
the solid models, animates the dynamics, displays the stresses,
and supports user manipulation and quantitative querying.
The computational requirements of the simulation module govern
the design of the other modules. We organize the requirements
around the tasks of reconstruction, kinematics, rigid body
dynamics, stress analysis, and flexible body dynamics.
