A statistical shape and alignment model was created for the structures of the knee: the femur, tibia and patella, associated articular cartilage, and soft tissue structures for a training set of 50 subjects/specimens. The structures of the knee were segmented from magnetic resonance images and an iterative closest point algorithm established nodal correspondence between a fine mesh for each member of the training set and a template mesh. Each of the structures was described in their local coordinate system and a 4x4 transformation was used to describe relative alignment between the structures in the as-scanned position. The statistical model utilized the nodal coordinates for the knee structures and the transformation matrices in a principal component analysis to capture the shape and alignment variability.

The statistical model describes intersubject anatomic variability in the shape and alignment of the knee structures and provides the ability to automatedly generate the geometry for a joint-level finite element analysis for members of the training set or virtual subjects derived from the statistical model, thus facilitating population-based evaluations.

As part of the project, we also prepared Biomechanics Education Modules that are targeted at middle and high school students (or even undergraduates). Please visit https://simtk.org/home/biomech_ed/ for more information.

This work was supported by the National Science Foundation, General and Age Related Disabilities Engineering under CBET Grant: 1034251.
"Population-based evaluation of knee mechanics considering inter-subject and surgical alignment variability"
Investigators: P. Laz, P. Rullkoetter, D. Dennis, R. Kim