SUNConferences, COMA '13

Font Size: 
Reverse Engineering the Human Knee
Johan van der Merwe, Cornie Scheffer, Dawie van der Heever, Pieter Erasmus

Last modified: 2013-08-21


This paper presents a method for reverse engineering human femoral cartilage. It forms part of an ongoing study in which healthy geometry for pathological knees is estimated using neural networks. The main characteristic of the method is that it delimits a well defined region of interest (ROI) on the cartilage surface, while preserving sufficient flexibility to accurately represent all geometric variations within the boundaries. The ROI is defined centrally to each condyle, and its width is based on a fraction of the medial-lateral width of the femur. The flexion axis is estimated by fitting two spheres to the posterior portions of the condyles. The resulting sphere radii for a single case compared well with literature. The parameterisation is a B-spline surface fitted globally in the least squares sense. For a flexion range corresponding to the normal gait cycle, errors between the parameterisation and segmented cartilage were found to be negligible.

Full Text: PDF