230
Biomedical Signal and Image Processing
The timing of the depolarization spikes in EMG signal in reference to the neural
impulses is used to identify the conduction and time response of muscles, as discussed earlier. Signal processing can be used to detect these peaks and measure the
time distance among these peaks. The information obtained from event timing studies will identify muscle conduction and general muscle health issues.
The force generated by a muscle directly affects the amplitude and frequency of
the EMG signal. Just like in measuring fatigue, different signal processing measures,
especially the root-mean-square (RMS) value, can be used to quantitatively measure
the muscle activity for such studies.
The ratio between muscle force and muscle contraction is an important measure
in evaluating the muscle condition and is often referred to as muscle’s mechanical
resistance. In calculating resistance, while muscle contraction is often measured by
direct measurement of the length of the muscle, muscle force is sometimes replaced
by the power of EMG as a more quantitative measure of force.
In surgery, where local anesthesia has to be performed, the anesthesiologist needs
to have a measure of how deep the local anesthesia is. The muscular response to
electric stimuli can help determine the level and depth of anesthesia and identify the
need to administer additional anesthetic drugs. In many cases, anesthesiologists use
the power of EEG (windowed over a period of at least several minutes) as the indicator of anesthetic efficacy.
Under certain conditions, the EMG of the contractions during childbirth is used
as a feedback mechanism to help the mother concentrate on the contractions. In this
case, the EMG is converted into an audible signal to make the onset of the contraction more noticeable.
11.6 PROCESSING AND FEATURE EXTRACTION OF EMG
After discussing the origin of the EMG signal and its applications in biomedical
diagnostics, next we focus on the typical processing methods applied for filtering
and analysis of EMG.
11.6.1 SOURCES OF NOISE ON EMG
In order to achieve two-way motion, for example, extending and flexing an arm, a
minimum of two muscle groups will be needed. Every skeletal muscle has an antagonist to stretch the contracted muscle after relaxation. Since each muscle can only
perform a one-way motion, it receives and generates only a binary signal, which can
be either on (i.e., contraction) or off (i.e., no contraction).
This agonist–antagonist principle is essentially the main source of cross talk and
noise in EMG, as it is almost always difficult to record the surface EMG of the agonist muscle without recording some of the electric activities of the antagonist muscle.
More specifically, due to the low resistance of the body around muscle cells, the
surface electrodes will detect action potentials of several muscle groups including
the antagonist muscles simultaneously.
This cross-talk effect can be avoided, at least to some extent, by filtering the
recorded signals and applying threshold detection to eliminate the signals from
Biomedical Signal and Image Processing
The timing of the depolarization spikes in EMG signal in reference to the neural
impulses is used to identify the conduction and time response of muscles, as discussed earlier. Signal processing can be used to detect these peaks and measure the
time distance among these peaks. The information obtained from event timing studies will identify muscle conduction and general muscle health issues.
The force generated by a muscle directly affects the amplitude and frequency of
the EMG signal. Just like in measuring fatigue, different signal processing measures,
especially the root-mean-square (RMS) value, can be used to quantitatively measure
the muscle activity for such studies.
The ratio between muscle force and muscle contraction is an important measure
in evaluating the muscle condition and is often referred to as muscle’s mechanical
resistance. In calculating resistance, while muscle contraction is often measured by
direct measurement of the length of the muscle, muscle force is sometimes replaced
by the power of EMG as a more quantitative measure of force.
In surgery, where local anesthesia has to be performed, the anesthesiologist needs
to have a measure of how deep the local anesthesia is. The muscular response to
electric stimuli can help determine the level and depth of anesthesia and identify the
need to administer additional anesthetic drugs. In many cases, anesthesiologists use
the power of EEG (windowed over a period of at least several minutes) as the indicator of anesthetic efficacy.
Under certain conditions, the EMG of the contractions during childbirth is used
as a feedback mechanism to help the mother concentrate on the contractions. In this
case, the EMG is converted into an audible signal to make the onset of the contraction more noticeable.
11.6 PROCESSING AND FEATURE EXTRACTION OF EMG
After discussing the origin of the EMG signal and its applications in biomedical
diagnostics, next we focus on the typical processing methods applied for filtering
and analysis of EMG.
11.6.1 SOURCES OF NOISE ON EMG
In order to achieve two-way motion, for example, extending and flexing an arm, a
minimum of two muscle groups will be needed. Every skeletal muscle has an antagonist to stretch the contracted muscle after relaxation. Since each muscle can only
perform a one-way motion, it receives and generates only a binary signal, which can
be either on (i.e., contraction) or off (i.e., no contraction).
This agonist–antagonist principle is essentially the main source of cross talk and
noise in EMG, as it is almost always difficult to record the surface EMG of the agonist muscle without recording some of the electric activities of the antagonist muscle.
More specifically, due to the low resistance of the body around muscle cells, the
surface electrodes will detect action potentials of several muscle groups including
the antagonist muscles simultaneously.
This cross-talk effect can be avoided, at least to some extent, by filtering the
recorded signals and applying threshold detection to eliminate the signals from