Page 317 - Biomedical Engineering and Design Handbook Volume 1, Fundamentals

294 BIOMECHANICS OF THE HUMAN BODY

12.3 PROCESSING THE EMG SIGNAL

The EMG signal detected by electrodes is a complicated signal, which is affected by the timing and
intensity of muscle contraction, the recording positions of electrodes, the quality of the contact between
electrodes and skin, the electrode and amplifier properties, the lab environment, etc. Here we will intro-
duce various procedures that can be used to analyze and interpret the EMG signal for different purposes.

12.3.1 Rectification
The EMG signal comprises positive and negative phases that fluctuate about a central line of zero
voltage (isoelectric line). Direct averaging the signal will not provide any useful information because
of the fluctuation about zero value. Thus, rectification is a necessary process for data analysis by
taking the absolute value of the EMG signal, that is, inverting the negative phases (Fig. 12.6).

t = 0 t = 0

Raw EMG After rectification After filtering
e(t)
FIGURE 12.6 Rectification and filtering of raw EMG. [Adapted from
Buchanan et al. (2004).]

12.3.2 Smoothing

The rectified signal still shows an irregular pattern due to different types of noises, so smoothing
should then be implemented to remove unwanted noises and get the desired information about
muscle activity (Fig. 12.7). High-pass filter and low-pass filter are two typical approaches to reduce
the low-frequency and high-frequency contents of the noise.
Movement of the electrodes may introduce motion artifact that is composed mostly of low-
frequency noise. Amplifiers of low quality may also introduce some low-frequency noise. It can be
corrected by high-pass filtering the EMG signal to eliminate these noises. The cutoff frequency
should be in the range of 5 to 30 Hz, depending on the type of filter and electrodes used (Buchanan
et al., 2004). This filter can be implemented in software, and a filter that has zero-phase delay properties
(e.g., forward and reverse pass fourth-order Butterworth filter) should be used, such that the filtering
does not shift the EMG signal in time. A high-pass filter with cutoff frequency of 5 to 30 Hz will

t = 0 t = 0 t = 0 t = 0

Raw EMG After rectification After filtering After activation After non-
e(t) dynamics u(t) linearization a(t)
FIGURE 12.7 Muscle activation dynamics: transformation from EMG to muscle activation. [Adapted from Buchanan
et al. (2004).]