【作者】 南文雅；

【导师】 明东；

【作者基本信息】 天津大学 ， 生物医学工程， 2009， 硕士

【摘要】 随着康复工程技术的发展,结合矫形器与步行器的矫形辅具在各类下肢运动功能障碍的治疗与康复中发挥出了越来越重要的作用。矫形辅具的助行效果会因其对关节约束模态条件的不同而有很大差别。如何在不同约束模态下准确全面地分析评估其助行效果已经成为指导矫形辅具有效使用的急待解决的研究课题。目前的步态分析技术更多关注的是生理学及运动学效果,缺乏稳定性特征对动作状态的评价和回馈,尤其是缺乏对包含全部下肢动作信息的支撑相稳定性特征的深入研究,因此往往使矫形辅具难以达到理想助行效果,这已成为困扰矫形辅具发展的瓶颈问题。本文分别从空间稳定性和力学稳定性的角度研究了矫形辅具助行过程中支撑相内的四种膝关节约束模态特征,旨在找出可表征不同矫形模式稳定性的敏感特征。研究中首先构建了基于步行器的无线测力系统,并利用传统线性方法和偏最小二乘回归方法对测力系统分别进行了静态标定,有关误差检验结果表明,偏最小二乘回归方法可有效降低测量中的交叉干扰,其最大交叉干扰为8.2%,远低于传统线性方法的19.96%。通过对15名受试者的膝部矫形辅具四模态助行实验,获得相关以柄反作用矢量表征的上肢三维力学信息,并进一步转化成单位步态周期支撑相内的步行器倾翻指数(WTI)和人体虚拟重心(VCG),分别以WTI和VCG作为力学稳定性和空间稳定性的特征向量,采用两种方法对四矫形模态进行分类:其一是利用判别分析法进行模态判别,其最高正确判别率可达93.33%;其二是用支持向量机方法进行模式分类,其最高识别率达91.67%。通过对判别分析和支持向量机方法的分析结果进行比较,证实这两种方法均可有效甄别出稳定性在支撑相内的矫形模态特征,但对于不同的稳定性特征会具有不同的敏感性。本研究结果表明,通过可靠的检测和分析方法,支撑相内的VCG与WTI可以作为区分不同矫形模态的敏感特征,用于针对性的康复指标制定,全面评价矫形辅具助行的步态效果,并有望在未来某些关节约束疾病的早期辅助检测和诊断中得到应用。更多还原

【Abstract】 With the development of rehabilitation engineering, orthosis assistive device combining orthosis and walker has played an important role in treatment and rehabilitation for dyskinesia of lower limbs. The assisted walking effects with orthosis assistive device vary obviously with the difference in joint constraint modal condition. How to analyze and assess these assisted walking effects accurately and comprehensively under different constraint modal conditions is the problem that should be solved as soon as possible. At present, gait analysis technology paid more attention on physiology and kinematics effect and lacked assessment and feedback of movement states with stability features, especially the stability features in stancephase which include all movement information of lower limbs. This deficit made orthosis assistive devices hardly to meet ideal assisted walking effect, which has been the bottleneck to the development of orthosis assistive devices.This thesis studied the spatial and mechanical stability features from four knee joint constraint modals during walking assisted with orthosis assistive device in order to find out those sensitive features which could describe stability in different orthopedic modals. Firstly, a wireless dynamometer system of walker was constructed and calibrated with traditional linear method and partial least-squares regression method. The result of error check suggested that the largest cross talk of partial least-squares regression method was 8.2%, which was much lower than 19.96% from traditional linear method. From the experiments of four modal assisted walking with orthosis assistive device of 15 subjects, three-dimensional mechanical information of upper limbs characterized by handle reaction vector were collected and converted into Walker Tipping Index (WTI) and Virtual Center of Gravity (VCG) in stance phase of gait cycle. WTI and VCG were separately used as feature vector of spatial and mechanical stability to classify four orthopedic models by two methods. One was discriminant analysi with the biggest recognition rate of 93.33%. Another was support vector machine with the biggest recognition of 91.67%. After comparing the results from discriminant analysis and support vector machine, it was suggested that both methods could identify orthopedic modal features of stability in satnce phase, but their sensitivity to various stability features was different.According to the results in this thesis, with reliable detection and analysis methods, VCG and WTI in stance phase can be sensitive features to distinguish different orthopedic models, which will be used to make targeted rehabilitation protocol and comprehensively assess the gait effect of walking assisted with orthosis assistive device. It will also likely to be applied in early aid detection and diagnosis in some joint constraint disease in the future.更多还原