基于样本熵和个体化阈值的表面膈肌肌电信号起点检测

Onset detection of surface diaphragmatic electromyography based on sample entropy and individualized threshold

膈肌是人体最主要的呼吸肌,表面膈肌肌电(sEMGdi)信号的动作区间起点检测可用于呼吸康复训练,但心电(ECG)信号的存在增加了其检测难度,故本文对此提出了基于样本熵(SampEn)和个体化阈值的起点检测方法,简称样本熵法。该方法涉及样本熵特征的提取,样本熵特征参数w和r0的优化,个体化阈值的选取以及判断条件的设立。同时还选用其他三种常用方法与本文所提的样本熵法进行起点检测方面的比较,即利用小波变换(WT)去噪后再分别使用均方根(RMS)和能量算子(TKE)的起点检测方法,以及不做小波变换而直接使用TKE的起点检测方法。本文共采集12名健康受试者在2种呼吸状态下的sEMGdi信号,用于信号合成和算法检测。最后以误差的绝对值累加和作为评价起点检测精度的指标。最终结果表明,样本熵法在稳定性和精度两方面皆优于其他三种方法,是一种能适应个体间差异,无需提前对sEMGdi信号进行ECG信号去噪便可获得较高精度的起点检测方法,为基于sEMGdi信号的呼吸康复训练和实时交互提供了依据。

The diaphragm is the main respiratory muscle in the body.The onset detection of the surface diaphragmatic electromyography (sEMGdi)can be used in the respiratory rehabilitation training of the hemiparetic stroke patients,but the existence of electrocardiography (ECG)increases the difficulty of onset detection.Therefore,a method based on sample entropy (SampEn)and individualized threshold,referred to as SampEn method,was proposed to detect onset of muscle activity in this paper,which involved the extraction of SampEn features,the optimization of the SampEn parameters w and τ0,the selection of individualized threshold and the establishment of the judgment conditions.In this paper,three methods were used to compare onset detection accuracy with the SampEn method,which contained root mean square (RMS)with wavelet transform (WT),Teager-Kaiser energy operator (TKE)with wavelet transform and TKE without wavelet transform,respectively,sEMGdi signals of 12healthy subjects in 2different breathing ways were collected for signal synthesis and methods detection.The cumulative sum of the absolute value of error T was used as an judgement value to evaluate the accuracy of the four methods.The results show that SampEn method can achieve higher and more stable detection precision than the other three methods,which is an onset detection method that can adapt to individual differences and achieve high detection accuracy without ECG denoising,providing a basis for sEMGdi based respiratory rehabilitation training and real time interaction.