INTELLIGENT ALGORITHMS FOR TRACKING THREE-DIMENSIONAL SKELETAL MOVEMENT FROM RADIOGRAPHIC IMAGE SEQUENCES

摘要

Systems and methods are disclosed that improve conventional tracking and modeling methods for determining three-dimensional bone motion from sequences of radiographic images. These enhancements significantly improve the speed, reliability and accuracy for bone motion tracking. Techniques used in various embodiments include multi-bone hierarchical techniques, time coherent approaches, simultaneous optimization of the entire motion sequence and improved initial estimates of bone motion paths.

主权项

1. A system that facilitates modeling a motion of a bone, comprising:
a processor operatively coupled to a memory, the memory comprising:
a three dimensional (3D) model of the bone and the one or more neighboring bones; anda sequence of frames of the motion of the bone, wherein each frame comprises one or more X-ray images of the bone and one or more neighboring bones; a multi-bone digitally reconstructed radiograph (MDRR) generation component that generates one or more MDRRs corresponding to the motion of the bone, wherein each MDRR is a two-dimensional projection of the 3D model of the bone and the one or more neighboring bones based at least in part on one or more motion parameters, wherein the one or more motion parameters comprise an estimated bone position and orientation; and an optimization component that determines a set of optimal motion parameters for the sequence based on a comparison between the one or more MDRRs with the one or more X-ray images of at least one frame according to an objective function, wherein the one or more optimal motion parameters are the one or more motion parameters when the comparison maximizes the objective function, and wherein, when the comparison does not maximize the objective function, the optimization component adjusts the one or more motion parameters, the MDRR generation component generates one or more additional MDRRs based at least in part on the one or more adjusted parameters, and the optimization component compares the one or more additional MDRRs with the one or more X-ray images of the at least one frame according to the objective function.