王建平,潘智光,王彦清,等.基于RSA的膝关节运动测量轮廓线误差分析[J].河南理工大学学报（自然科学版）,2026,45(6):108-116.

Wang J P, Pan Z G, Wang Y Q, et al.Error analysis of knee joint motion measurement contour based on RSA[J].Journal of Henan Polytechnic University(Natural Science) ,2026,45(6):108-116.

基于RSA的膝关节运动测量轮廓线误差分析

王建平1, 潘智光1, 王彦清2, 邵培轩3, 马素霞4, 张阳5, 张明军1, 宋梦杰6, 陈旭7, 郭栋8, 曾西9, 谈建新10, 胡海11

1.河南理工大学 机械与动力工程学院，河南 焦作 454003；2.苏州大学 放射医学与防护学院，江苏 苏州  215006；3.无锡学院 物联网工程学院，江苏 无锡  214105；4.焦作市第二人民医院 口腔科，河南 焦作  454003；5.上海旗澜实业有限公司，上海  201306；6.焦作市新港医疗设备有限公司，河南 焦作  454000；7.焦作市第二人民医院 骨科，河南 焦作  454003；8.河南理工大学（鹤壁校区） 智能制造学院，河南 鹤壁  458000；9.郑州大学第一附属医院 康复医学科，河南 郑州  450052；10.焦作市中医院 康复科，河南 焦作  454000；11.上海市第六人民医院 骨科，上海  200233

摘要: 目的 通过放射立体测量分析（radiostereometric analysis，RSA）技术对膝关节运动测量中的轮廓线误差进行系统分析，明确误差的大小、来源，并提出有效的误差减小方法，以优化膝关节运动测量流程，提高测量精度，为膝关节生物力学性能的准确评估提供理论支持与实践指导。  方法 选取30位健康志愿者，对其膝关节进行CT扫描以建立三维骨组织模型，并利用C型臂X射线机采集动态透视图像。6位工作人员在Rhino软件中手动绘制轮廓线，并通过多相位配准技术将三维模型与轮廓线对齐。采用分割逼近法结合MATLAB中的fminunc优化函数进行循环迭代计算，得到轮廓线误差。分析图像质量、工作人员操作熟练度、CT扫描数据精度以及膝关节三维模型复杂性等因素对误差的影响，提出相应的优化方法。  结果 研究结果显示，6位工作人员绘制的股骨轮廓线上测量点到理论轮廓线法向距离的误差均值为0.960 mm，胫骨轮廓线误差均值为0.896 mm。探讨图像质量、工作人员操作熟练度、CT扫描数据精度以及膝关节三维模型复杂性等误差来源，提出通过图像灰度及对比度处理等方法减小误差。 结论 研究结果可为膝关节运动测量流程的优化提供理论支持与实践指导，有助于更准确地评估膝关节生物力学性能，为相关医学研究和临床应用提供参考。

关键词:RSA技术;膝关节运动测量;轮廓线误差;分割逼近法;2D-3D配准;误差来源分析

doi:10.16186/j.cnki.1673-9787.2025030004

基金项目:国家自然科学基金资助项目（31370999）；中国残疾人联合会残疾人辅助器具专项项目（2024CDPFAT-15）；河南省科技攻关项目（232102311185，262102311225）

收稿日期:2025/03/03

修回日期:2025/05/23

出版日期:2026/10/09

Error analysis of knee joint motion measurement contour based on RSA

Wang Jianping1, Pan Zhiguang1, Wang Yanqing2, Shao Peixuan3, Ma Suxia4, Zhang Yang5, Zhang Mingjun1, Song Mengjie6, Chen Xu7, Guo Dong8, Zeng Xi9, Tan Jianxin10, Hu Hai11

1.School of Mechanical and Power Engineering， Henan Polytechnic University， Jiaozuo  454003， Henan， China；2.School of Radiation Medicine and Protection， Soochow University， Suzhou  215006， Jiangsu， China；3.School of Internet of Things Engineering， Wuxi University， Wuxi  214105， Jiangsu， China；4.Department of Stomatology， The Second People’s Hospital of Jiaozuo， Jiaozuo  454003， Henan， China；5.Shanghai Qilan Industrial Co.， Ltd.， Shanghai  201306， China；6.Xin’gang Medical Equipment Co.， Ltd.， Jiaozuo  454000， Henan， China；7.Department of Orthopedics， The Second People’s Hospital of Jiaozuo， Jiaozuo  454003， Henan， China；8.School of Intelligent Manufacturing， Hebi Campus， Henan Polytechnic University， Hebi  458000， Henan， China；9.Department of Rehabilitation Medicine， The First Affiliated Hospital of Zhengzhou University， Zhengzhou  450052， Henan， China；10.Department of Rehabilitation Medicine， Jiaozuo Traditional Chinese Medicine Hospital， Jiaozuo  454000， Henan， China；11.Department of Orthopedics， Shanghai Sixth People’s Hospital， Shanghai  200233， China

Abstract: Objectives To systematically analyze the contour errors in knee joint motion measurement using radiostereometric analysis （RSA） technology， clarify the magnitude and sources of these errors， and propose effective error reduction methods. The aim is to optimize the knee joint motion measurement workflow， improve measurement accuracy， and provide both theoretical support and practical guidance for the accurate assessment of knee joint biomechanical performance.  Methods Thirty healthy volunteers were selected. CT scans of their knee joints were performed to establish three-dimensional bone models， and dynamic fluoroscopic images were acquired using a C-arm X-ray machine. Six operators manually drew contour lines in Rhino software and aligned the 3D models with the contour lines through multi-phase registration techniques. A segmentation approximation method combined with the fminunc optimization function in MATLAB was employed for iterative computation to determine the contour errors. The effects of image quality， operator proficiency， CT data accuracy， and knee joint model complexity on the contour errors were analyzed， and corresponding optimization methods were proposed.  Results The results showed that the mean contour errors of the femoral and tibial contours drawn by the six operators were 0.960 mm and 0.896 mm， respectively， measured as the normal distance from the measurement points on the contour to the theoretical contour. Error sources including image quality， operator proficiency， CT data accuracy， and knee joint model complexity were analyzed， and methods such as image grayscale and contrast adjustment were proposed to reduce the errors.  Conclusions The findings provide both theoretical support and practical guidance for optimizing the knee joint motion measurement workflow， contributing to more accurate assessment of knee joint biomechanical performance and offering valuable references for related medical research and clinical applications.

Key words: RSA technology; knee joint motion measurement; contour line error; segmentation approximation method;2D-3D registration; error source analysis