摘要
目的利用RGB-D数据驱动骨肌模型来预测受试者步行时的膝关节接触力,并验证其准确性。方法首先,利用Kinect和Qualisys运动捕捉系统(参照标准)同步采集了6名受试者的运动轨迹数据;然后,利用运动轨迹数据和地面反作用力数据分别驱动校准后的骨骼肌肉模型,通过模型的逆向动力学分析功能实现了膝关节接触力的预测;最后,利用均方根误差(δRMSE)和皮尔逊相关系数(ρ)对两种方法的预测结果进行了对比验证。结果和参照方法相比,RGB-D数据驱动的骨肌模型能准确预测受试者步行时的膝关节接触力(δRMSE=77.1;ρ=0.970)。结论RGB-D数据驱动的骨肌模型能够精准预测膝关节接触力,可作为一种更好的替代方法在临床上应用。
Objective To predict the knee contact force during walking and to verify the accuracy of RGBD data-driven musculoskeletal model.Methods Kinect and Qualisys motion capture system(reference standard)were used to collect the motion trajectory data of 6 subjects synchronously.The calibrated musculoskeletal model was driven by the motion trajectory data and ground reaction force data respectively.Then the contact force of the knee joint was predicted by the reverse dynamics analysis function of the model.In the end,the root mean square error(δRMSE),and Pearson correlation coefficient(ρ)were used to compare and verify the predicted results of the two methods.Results Comparing with the reference method,the RGB-D data-driven musculoskeletal model could accurately predict the contact force of the knee joint during walking(δRMSE=77.1;ρ=0.970).Conclusion RGB-D data-driven musculoskeletal model can accurately predict the contact force of the knee joint,and can be used as a better alternative in clinical practice.
作者
徐唯祎
朱业安
王浩伦
卢巍
Xu Weiyi;Zhu Yean;Wang Haolun;Lu Wei(不详;Rehabilitation Medicine Department,Jiangxi Provincial Peoples Hospital,Nanchang Jiangxi 330006,China)
出处
《航天医学与医学工程》
CAS
CSCD
北大核心
2020年第3期252-257,共6页
Space Medicine & Medical Engineering
基金
中国高校产学研创新基金(用友专项)
江西省科技厅重点研发计划项目(20192BBG70011)。
