Veer Karan, Sharma Tanu
D.S. Kothari Postdoctoral Fellow (University Grant Commission), New Delhi, India.
Computer Science Engineering Department (CSED), Global College of Engineering and Technology, Khanpur Kui, Ropar, India.
Biomed Tech (Berl). 2018 Mar 28;63(2):131-137. doi: 10.1515/bmt-2016-0038.
Dual-channel evaluation of surface electromyogram (SEMG) signals acquired from amputee subjects using computational techniques for classification of arm motions is presented in this study. SEMG signals were classified by the neural network (NN) and interpretation was done using statistical techniques to extract the effectiveness of the recorded signals. From the results, it was observed that there exists a calculative difference in amplitude gain across different motions and that SEMG signals have great potential to classify arm motions. The outcomes indicated that the NN algorithm performs significantly better than other algorithms, with a classification rate (CR) of 96.40%. Analysis of variance (ANOVA) presents the results to validate the effectiveness of the recorded data to discriminate SEMG signals. The results are of significant thrust in identifying the operations that can be implemented for classifying upper-limb movements suitable for prostheses' design.
本研究提出了使用计算技术对截肢者受试者采集的表面肌电图（SEMG）信号进行双通道评估，以对手臂运动进行分类。通过神经网络（NN）对SEMG信号进行分类，并使用统计技术进行解释，以提取记录信号的有效性。从结果中可以观察到，不同运动的幅度增益存在计算差异，并且SEMG信号在分类手臂运动方面具有很大潜力。结果表明，NN算法的性能明显优于其他算法，分类率（CR）为96.40%。方差分析（ANOVA）给出结果以验证记录数据区分SEMG信号的有效性。这些结果对于确定可用于分类适合假肢设计的上肢运动的操作具有重要意义。
