Independent component analysis of mechanomyogram detected with an acceleration sensor in motion

Mechanomyography is a novel technique that reveals muscle activity and mechanical properties. A mechanomyogram is often measured with an acceleration sensor. The signal recorded consists of both the mechanomyogram andmotion acceleration. The purpose of this study is to separate the mechanomyogram from the mixed signal measured with an acceleration sensor. Five healthy male subjects participated in the experiment. Each subject sat on a chair, his arm fixed on a smoothly rotating, horizontal disk. Three acceleration sensors were attached to the subject's forearm. The subject was instructed to flex and extend his elbow while tracking a target displayed on a screen. Acceleration was measured during elbow flexion and extension. The independent component analysis was then applied to the recorded signal, and the separated signals were compared to the reference signals. The separated signal corresponding to the mechanomyogram was full-wave rectified and integrated to obtain the time course of the amplitude (IMMG). The reference signal of the mechanomyogram was the second derivative of the displacement mechanomyogram measured with a condenser microphone that is not affected by the motion acceleration. The reference signal of the motion was the second derivative of the elbow angle measured with a goniometer. The mechanomyogram and motion acceleration could be separated by the proposed method. The IMMG agreed with the rectified smoothed reference signal, while the mean power frequency of the separated signal did not agree with that of the reference signal.