基于经验模态分解的膝骨性关节炎步态分类  

Classification of Knee Osteoarthritic Gait Based on Empirical Mode Decomposition

作　　者：黄春耀 曾玮[2] HUANG Chunyao;ZENG Wei(Longyan Technician College,Longyan,Fujian 364000,China;Longyan University,Longyan,Fujian 364000,China)

机构地区：[1]龙岩技师学院电气工程系,福建龙岩364000 [2]龙岩学院物理与机电工程学院,福建龙岩364000

出　　处：《龙岩学院学报》2020年第5期29-35,共7页Journal of Longyan University

基　　金：国家自然科学基金(61773194);福建省自然科学基金(2018J01542);龙岩技师学院产学研项目(LYJS2020A02)。

摘　　要：提出了一种新颖的基于经验模态分解(Empirical Mode Decomposition,EMD)和神经网络的膝骨性关节炎步态模式特征提取与分类的方法,可用于区分病理性步态和正常步态。对19例膝骨性关节炎患者和28例作为对照组的正常人进行实验研究,并采用2-折(Two-Fold)和留一(Leave-One-Out)交叉验证法可以分别达到91.3%和93.6%的分类准确度。与其他最新的方法相比较,该分类方法能有效地区分作为对照者的正常人和膝骨性关节炎患者之间的步态模式,具有较好的分类效果和显著的有效性。We propose a novel method for feature extraction and classification of knee gait patterns based on empirical mode decomposition(EMD)and neural networks,and it can be used to discriminate pathological gait from normal gait.Experiments are carried out on 19 patients with knee OA and 28 normal people as a control group,and the accuracy of classification is 91.3%and 93.6%respectively by two fold and leave one out cross validation.Compared with other state-of-the-art algorithms,this method can effectively distinguish the gait patterns between healthy controls and patients with knee osteoarthritis,which demonstrates a better classification performance and significant effectiveness.

关 键 词：膝骨性关节炎步态识别 步态模式特征建模 经验模态分解 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]