Dakin Christopher J, Dalton Brian H, Luu Billy L, Blouin Jean-Sébastien
School of Kinesiology, University of British Columbia, Vancouver, British Columbia, Canada;
School of Kinesiology, University of British Columbia, Vancouver, British Columbia, Canada; Brain Research Center, University of British Columbia, Vancouver, British Columbia, Canada; and Institute for Computing, Information and Cognitive Systems, University of British Columbia, Vancouver, British Columbia, Canada
J Neurophysiol. 2014 Oct 1;112(7):1685-91. doi: 10.1152/jn.00296.2014. Epub 2014 Jul 2.
Rectification of surface electromyographic (EMG) recordings prior to their correlation with other signals is a widely used form of preprocessing. Recently this practice has come into question, elevating the subject of EMG rectification to a topic of much debate. Proponents for rectifying suggest it accentuates the EMG spike timing information, whereas opponents indicate it is unnecessary and its nonlinear distortion of data is potentially destructive. Here we examine the necessity of rectification on the extraction of muscle responses, but for the first time using a known oscillatory input to the muscle in the form of electrical vestibular stimulation. Participants were exposed to sinusoidal vestibular stimuli while surface and intramuscular EMG were recorded from the left medial gastrocnemius. We compared the unrectified and rectified surface EMG to single motor units to determine which method best identified stimulus-EMG coherence and phase at the single-motor unit level. Surface EMG modulation at the stimulus frequency was obvious in the unrectified surface EMG. However, this modulation was not identified by the fast Fourier transform, and therefore stimulus coherence with the unrectified EMG signal failed to capture this covariance. Both the rectified surface EMG and single motor units displayed significant coherence over the entire stimulus bandwidth (1-20 Hz). Furthermore, the stimulus-phase relationship for the rectified EMG and motor units shared a moderate correlation (r = 0.56). These data indicate that rectification of surface EMG is a necessary step to extract EMG envelope modulation due to motor unit entrainment to a known stimulus.
在将表面肌电图（EMG）记录与其他信号进行关联之前进行校正，是一种广泛使用的预处理形式。最近，这种做法受到了质疑，使得EMG校正这一主题成为了一个备受争议的话题。支持校正的人认为，它突出了EMG尖峰定时信息，而反对者则指出这是不必要的，并且其对数据的非线性失真可能具有破坏性。在这里，我们研究了校正对于提取肌肉反应的必要性，但首次使用电前庭刺激这种已知的肌肉振荡输入形式。参与者在接受正弦前庭刺激的同时，记录左侧腓肠肌内侧的表面和肌内EMG。我们将未经校正和校正后的表面EMG与单个运动单位进行比较，以确定哪种方法能在单个运动单位水平上最好地识别刺激-EMG相干性和相位。在未经校正的表面EMG中，刺激频率下的表面EMG调制很明显。然而，这种调制未被快速傅里叶变换识别出来，因此与未经校正的EMG信号的刺激相干性未能捕捉到这种协方差。校正后的表面EMG和单个运动单位在整个刺激带宽（1 - 20 Hz）上均显示出显著的相干性。此外，校正后的EMG与运动单位的刺激相位关系具有中等程度的相关性（r = 0.56）。这些数据表明，对表面EMG进行校正是提取由于运动单位对已知刺激的同步化而产生的EMG包络调制的必要步骤。
