Optimizing image similarity metrics for model-based biplanar fluoroscopy foot and ankle bone tracking

Model-based biplanar fluoroscopic bone tracking is a powerful tool for biomechanists seeking to quantify in vivo, functional joint motion. The sizes, shapes, and configurations of the osseous foot and ankle anatomy present a series of unique challenges to obtaining optimal kinematic data with minimal irradiation of test subjects. We sought an experimental approach to optimizing the 3D-to-2D model-based registration algorithms in our custom bone tracking software suite.