利用AR模型参数和BP神经网络辨识损伤神经的针电极肌电信号

Using ARmodel and BP ANN to identify injurious nerve with node EMG signal

在临床中,通过观察异常募集的肌电信号可以用来判断肌神经的损伤程度。异常募集的主要特征值包括MUP的幅度、时限、发放频率、单位时间的转折数等。本文作者提出了一种对前臂肌神经损伤肌肉的针电极肌电信号建立AR模型,提取AR系数并通过BP神经网络判别肌神经损伤的方法。通过实验证明,这种方法判别准确度高、速度快,在计算机辅助诊疗肌神经损伤中有着良好的应用前景。

Generally,abnormal recruitment NEMG is usually used to evaluate the injurious degree of the nerve in clinic.The eigenvalues of the abnormal recruitment signal include the amplitude of the MUP,duration,recruitment frequency,number of the turns per unit time and so on.In this paper,a new method to determine the injury of the muscle nerve such as neurotmesis and neuropathy is advanced.This method creates a AR model for the forearm NEMG signal and uses the BP artificial neural network to identify the muscle nerve injury with AR model coeffi-cient.Compared with the traditional methods,this measure has the advantage in accuracy of the identified rate and the speed.So it will have a wide application prospects in the computer-aided diagnosis.