Toward a More Efficient and Effective Method for Tracking 3D Bone Position and Orientation in Fluoroscopic Images

Osteoarthritis (OA) is the primary cause of mobility-based disability in the United States. Approximately 12% of osteoarthritis cases occur following traumatic injury, through a process termed post-traumatic osteoarthritis (PTOA). PTOA accelerates the degeneration of articular cartilage, thus having profound effects, particularly on the young athletic population. It has been found that, following injury, there are both biological changes to the articulating cartilage as well as dynamic changes to the knee joint kinematics. However, it remains unclear how these changes in kinematics correlate with cartilage degradation. Understanding this relationship of the mechanisms that lead to PTOA assists in the development of effective treatments for patients who have suffered previous knee injuries. In order for researchers to be best equipped to observe and understand how altered kinematics following surgery result in cartilage degradation, new methods must be developed to quantify these changes to the kinematics. Two of the leading methods for measuring bone movement and studying knee joint mechanics are motion capture and dual fluoroscopy with model-based tracking (DF-MBT). The primary problems with these methods are that motion capture, while providing near immediate results, is relatively inaccurate since it suffers from soft tissue artifact. DF-MBT, on the other hand, is considered accurate to within a millimeter, but proper post-processing of the data requires a significant amount of time. Ideally, there would be a method that provided results as accurately as the DF-MBT method in the time interval of the motion capture method. Therefore, the purpose of this study was to develop a tracking method that is faster than DF-MBT and accurate on the order of a millimeter or less. The initial methodology behind achieving this goal was to write a code that would produce a transformation matrix between the DF-MBT and motion capture pose maps. A pose map is a representation of the position (pose) of an object in reference to a particular coordinate system at each frame. Therefore, a static frame could be used in order to define a transformation from motion capture clusters to the bone\u27s location. This transformation matrix can then be applied to the dynamic frames to produce results on the same level of accuracy in a much shorter time frame. This study analyzed and compared the results of three different methods of measuring bone movement; traditional DF-MBT tracking (Model Based Tracking), the previously described pose map method (Skin Marker Tracking), and an additional method which combined the process of the other two methods (Combined Tracking). Despite the Skin Marker Method providing unsatisfactory results, the goal of producing a more efficient and effective method for measuring bone movement was still achieved. The Combined Tracking method, which involved using the transformation code as a starting point for traditional DF-MBT alignment, resulted in more accurate results than DF-MBT in a shorter time frame. Therefore, a more efficient and effective method for tracking 3D bone position and orientation in fluoroscopic imaging was successfully developed