Electromyography (EMG)

The EMG circuit
The raw emg signal
Electrodes and signal detection
Signal processing and amplification
Outcome measures

NB: I created this page in 1998 for graduate physical therapy students. Visitors should note that the information dates from that time. I apologize that I do not currently work in the field, and cannot provide updated information. I've left the page here because some have told me that the overview of EMG, though a little dated, is a useful introduction. David M. Thompson, PT

The circuit:

The raw EMG signal:

a voltage difference or difference in electrical potential (E=Ir) measured between recording electrodes The signal's origins include electrical activity in various tissues: potentials in motor units of muscle fibers (muap)

motor neuron action potential produces an end plate potential

end plate potential directly depolarizes muscle membrane near muscle fiber's center

depolarization wave travels from muscle's surface along transverse tubules into fiber's interior

depolarization travels simultaneously along membrane's surface by local current flow, creating a train of bipoles +/-.

The signal is detected by electrodes:
surface (Ag-AgCl) indwelling
  • needle
  • fine wire
surface area large area behaves like a string of point electrodes; each 'point' picks up the same signal, slightly delayed in time; the sum is a longer duration waveform. small area picks up more discrete signals without producing long duration waveform
pick up zone limited to 0.5 cm to 1.5 cm; can record only from superficial muscles necessary for recording from deep muscles
"cross talk" (pickup of signals from adjacent muscles) more common less common
choice of electrode placement for electrode to detect a muap, it must travel in a direction such that the distance between impuse and electrode changes

Therefore, electrodes should be aligned with expected direction of impulse (or aligned perpendicular to impulses that should be excluded)

Signal processing and amplification

  1. rectification Because the raw signal is biphasic, its mean value is zero. The rectifier allows current flow in only one direction, and so "flips" the signal's negative content across the zero axis, making the whole signal positive.

  2. Filtering and linear envelope detection EMG is actually a composite of many signals, as well as some noise. These voltages also rise and fall at various rates or frequencies, forming a frequency spectrum. Circuits filter the composite signal and eliminate unwanted and meaningless electrical noise like movement artifact. Because most EMG exists in a frequency range between 20 and 200 Hz, and because movement artifacts have frequencies less than 10 Hz, that is the cutoff frequency frequently employed in "high-pass" filters. The

  3. Integration calculating the area under the linear envelope, a quantity analogous to electrical work or energy.

  4. Amplification single muaps have amplitude of 100 microvolts
    signals detected by surface electrodes are in range of 5 mV
    signals detected by indwelling electrodes are in range of 10 mV

    Amplifier impedance must be considerably higher than the impedance of the electrode/skin or electrode/muscle interfaces. Since indwelling electrodes have very high impedances (due to low surface area), very high amplifier impedances are necessary.

    common mode rejection

    Even when leads are shorted together, a signal of 20 - 50 microvolts may result from ambient electromagnetic energy, induced by nearby machinery or electrical lines. The body itself is like an antenna that picks up ambient electromagnetic signals, including the 60 Hz signal from ordinary house current. This "hum" is in the middle of EMG frequency range; we can't filter it.

Outcome measures:

vertical axis (voltage)
    rate of rise of voltage
    peak amplitude

horizontal axis (time)

Ellaway's cumulative sum histogram technique:

Where Si is the cumulative sum up to sample i, X is the mean voltage over the trial, and xi is the voltage at sample i.

Ellaway, P.H. (1978). Cumulative sum technique and its application to the analysis of peristimulus time histograms. Electroencephalography Clinical Neurophysiology, 45, 302-304.

Frequency analysis or spectral analysis

Because the emg signal is actually a summation of muap, some close and some distant from the recording electrodes, it is difficult to know which motor units contribute to it. Some investigators try to divide the signal's components into its constituent frequencies.

In fatigued motor units, firing rate and twitch tension decrease, while contraction time increases, perhaps because muap conduction is slower in fatigued muscle.

In more fatiguable, fast conducting and fast twitch (therefore high frequency) motor units, the system compensates for fatigue by recruiting new motor units. This produces more synchronous firing and a lower frequency bias in the composite EMG signal from fatigued muscle.

Reliability of EMG signal; is it reproduceable?

Validity of EMG signal: its relation to meaningful biomechanical variables: Do variations in EMG signal reflect real variations in

On and off times may be distorted by:

  1. movement artifact. We need to eliminate frequency spectra that are likely to be artifactual, using a smoothing or averaging technique like Ellaway's

  2. electromechanical delay Winter (p.184) explains electromechanical delay in terms of muscular anatomy. Because the muscle's contractile proteins develop force in parallel with elastic and viscous elements, muscle force must elongate these elastic and viscous proteins in their fluid environment before tension is registered in tendon. Electromechanical delay may vary inversely with the velocity of muscle shortening.

Normalized EMG is grossly associated with changes in muscle force. However, the relationship is most consistent and linear when:

  1. muscle action is isometric. When a muscle changes length, its moment arm also changes, and can produce a non-linear force-emg relationship. Soderberg and Cook (1984) found that EMG and force are linear during isometric actions at intermediate lengths.

    Studies of EMG during maximal isokinetic testing have showed that EMG may not change when the muscle's force changes. In particular, when performing equal magnitudes of negative (eccentric) work and positive (concentric) work, muscle's exhibit less EMG activity during the eccentric activity.

  2. muscle length is consistent between trials. At longer muscle lengths, researchers have recorded lower amplitude EMG even when the muscle's force is increased, perhaps because:

    • when the muscle is elongated, less muscle mass lies under recording electrodes

    • activated muscle spindles produce reflex inhibition of alpha motor neurons see Soderberg & Cook. (1984). Physical Therapy, 64, 1817.

  3. muscles increase force by increasing their motor neuron firing rate, not by recruiting additional motor units. Recruited motor units, because they are larger, lead to a non-linear changes in force-emg relationship.

EMG information last updated 3-2-98
page last updated 43-31-2006
David M. Thompson, PT
return to Pathokinesiology home page