【摘要】 本文旨在采用表面肌电信号无创性方法诊断和评判膝骨性关节炎,以在早期能够预防和治疗膝骨性关节炎,改善生活质量。在研究中,采集了对照组和膝骨性关节炎患者水平行走时下肢的股外侧肌,股内侧肌,股二头肌和半腱肌的表面肌电信号。利用表面肌电信号建立自回归(AR)模型,提取AR模型参数为特征向量训练BP神经网络,并通过神经网络诊断膝骨性关节炎。实验表明,基于BP神经网络分类器可以得到较好的结果,正确率可达到88%以上。更多还原

【Abstract】 It is aimed to diagnose and evaluate knee osteoarthritis with noninvasive method by surface electromyography (sEMG) signal, which can prevent and treatment knee osteoarthritis in early stages and improve the quality of life. In this study, the sEMG signals were collected, which from the vastus lateralis, vastus medialis, biceps femoris, and semitendinosus of lower extremity during level working among control subjects and knee osteoarthritis patients. An autoregressive (AR) model is built with sEMG. The AR model parameters are extracted as the characteristic vectors, which is used to train the BP neural network. Then the knee osteoarthritis is diagnosed through the BP neural network. It is showed from the experiments that a satisfactory result is achieved from classifiers based on BP neural network, with the accuracy rate more than 88%.更多还原