Sarojak M, Hoff W, Komistek R, Dennis D
Colorado School of Mines, 1500 Illinois Ave, Golden, CO 80401, USA.
Biomed Sci Instrum. 1999;35:9-14.
The objective of this project was to develop and test a software system for the kinematic analysis of total joint arthroplasty (TJA) implants. Using a supervised iterative optimization algorithm (simulated annealing), the system iteratively adjusts the pose of an implant model to maximize correlation between the model's pose and the pose of an actual implant in a x-ray image. A graphical user interface (GUI) was developed that provides visualization of the model-fitting process, and permits human guidance for error correction. The resulting system was evaluated to determine its accuracy and repeatability. On synthetic images, the mean translational and rotational errors were found to be 0.005 mm and 0.0015 degrees, respectively. On in vitro images, the repeatability was found to be 0.15 mm of translation and 0.17 degrees of rotation. The outcome is a working joint-measurement software system and quantitative data to support comparison with other methods.
本项目的目标是开发并测试一个用于全关节置换(TJA)植入物运动学分析的软件系统。该系统使用一种监督式迭代优化算法(模拟退火算法),迭代调整植入物模型的姿态,以最大化模型姿态与X射线图像中实际植入物姿态之间的相关性。开发了一个图形用户界面(GUI),用于可视化模型拟合过程,并允许人工指导进行误差校正。对所得系统进行了评估,以确定其准确性和可重复性。在合成图像上,平均平移误差和旋转误差分别为0.005毫米和0.0015度。在体外图像上,平移的可重复性为0.15毫米,旋转的可重复性为0.17度。成果是一个实用的关节测量软件系统以及支持与其他方法进行比较的定量数据。