Building Biomechanical Models Based on Medical Image Data: An Assessment of Model Accuracy

The goals of this work were to (i) establish a method for building subject-specific biomechanical models from medical image data, (ii) construct a subject-specific model of the elbow, and (iii) quantify the accuracy of soft tissue excursions estimated from the model. We developed a kinematic model of the elbow joint and its surrounding musculature from magnetic resonance images of a 6'4" male cadaver specimen in one limb position. Moment arms estimated from the model (i.e., the changes in muscle-tendon lengths with elbow flexion angle) were compared to moment arms measured experimentally from the same specimen. In five of the six muscles studied, the model explained 84%-94% of the variation in the experimental data. Model estimates of peak elbow flexion moment arm were within 13% of the experimental peaks. Our results suggest that subjectspecific musculoskeletal models derived from medical image data have the potential to substantially improve estimates of soft tissue excursions in living subjects.