Surface EMG Glossary

Signal origin and physiology#

Motor unit#

A motor unit is one alpha motor neuron together with all the muscle fibres it innervates; it is the smallest functional unit the nervous system can activate.

The number of fibres per unit ranges from a handful (eye muscles) to more than a thousand (large leg muscles). Everything a surface electrode records is the summed activity of many motor units.

Motor unit action potential (MUAP)#

A motor unit action potential is the summed electrical potential of all fibres of one motor unit firing once, as seen from a given electrode position.

Its shape depends on the fibre geometry, the depth of the unit and the tissue between fibres and electrode. Surface EMG never shows single MUAPs cleanly — it shows their interference pattern.

Interference pattern#

The interference pattern is the raw surface EMG signal that results from many motor units firing asynchronously and overlapping in time.

Because it is stochastic, two identical contractions never produce an identical raw trace. That is why amplitude is always analysed over a time window (see RMS), never at a single sample.

Recruitment and rate coding#

Recruitment is the activation of additional motor units, rate coding the increase of firing frequency of units already active; together they grade muscle force.

Both raise the surface EMG amplitude, which is why amplitude rises with force — but not linearly and not identically across muscles.

Innervation zone (motor point)#

The innervation zone is the region of a muscle where the motor axons terminate on the fibres — the origin of the action potentials, which then propagate in both directions towards the tendons.

Electrodes placed over the innervation zone pick up potentials travelling in opposite directions that partially cancel, giving unstable, smaller amplitudes. SENIAM therefore places electrodes between the innervation zone and the distal tendon.

Conduction velocity#

Muscle fibre conduction velocity is the speed at which an action potential travels along a muscle fibre, typically 3–5 m/s.

It falls with fatigue and rises with temperature; its change is the physiological reason the EMG frequency spectrum shifts during sustained contractions.

Amplitude cancellation#

Amplitude cancellation is the loss of signal that occurs when positive and negative phases of overlapping action potentials sum to zero at the electrode.

It is the main reason surface EMG amplitude underestimates true neural drive at high activation levels — and one reason placement over the innervation zone should be avoided.

Electrodes and placement#

Bipolar (differential) derivation#

A bipolar derivation records the voltage difference between two electrodes over the same muscle; signals common to both electrodes are cancelled.

This is the standard configuration in surface EMG. Interference from mains hum or the heart that reaches both electrodes almost equally is largely removed (see CMRR).

Inter-electrode distance (IED)#

The inter-electrode distance is the centre-to-centre distance between the two detection electrodes of a bipolar pair; SENIAM recommends 20 mm.

A smaller distance increases selectivity (less crosstalk) but reduces amplitude; a larger distance picks up more of the muscle and more of its neighbours. For very small muscles SENIAM allows a smaller IED, but never larger than a quarter of the fibre length.

Reference (ground) electrode#

The reference electrode provides the common electrical potential for the amplifier and is placed over electrically inactive tissue, typically a bony prominence.

Common sites are the wrist, the ankle, the C7 spinous process or the iliac crest. A missing or badly attached reference electrode is the first thing to check when the baseline is noisy.

Electrode orientation#

Electrode orientation is the direction of the line connecting the two detection electrodes relative to the muscle fibres; it should be parallel to the fibre direction.

Placing the pair across the fibres reduces amplitude and distorts the frequency content, because the two electrodes then no longer "see" the same travelling action potential one after the other.

Anatomical landmarks#

Anatomical landmarks are palpable bony points — acromion, epicondyles, spina iliaca, malleoli — from which SENIAM defines electrode positions as fractions of a connecting line.

Working with landmarks instead of "the middle of the muscle" is what makes a placement reproducible between sessions and between examiners.

Skin preparation#

Skin preparation is the removal of hair, dead skin cells and sweat before electrode application, to lower and stabilise skin impedance.

Shaving, light abrasion with an abrasive paste or fine sandpaper, then cleaning with alcohol; the skin should look slightly reddened. For slow, static tests alcohol alone may suffice; for running or jumping thorough preparation is mandatory.

Skin impedance#

Skin impedance is the electrical resistance the electrode-skin interface opposes to the current; it should be low and similar under both electrodes.

Konrad's practical scale: 1–5 kΩ very good, 5–10 kΩ good, 10–30 kΩ acceptable for easy conditions, above 50 kΩ repeat the preparation. Impedance falls by more than half during the first minutes after application, so wait before measuring.

Ag/AgCl electrode#

A silver/silver-chloride electrode is the standard pre-gelled disposable surface electrode; its non-polarisable interface produces a stable, low-noise contact.

SENIAM recommends a conductive area of 10 mm or less in diameter. Wet-gel types give lower impedance; adhesive-gel types can be repositioned.

Crosstalk#

Crosstalk is the portion of a surface EMG signal that originates from muscles other than the one under the electrode.

It cannot be recognised in the raw trace. It is minimised by small electrodes, a short inter-electrode distance, placement over the muscle belly away from borders, and by choosing muscles that are actually superficial. Every muscle page on EMG Guide lists the typical crosstalk neighbours.

Volume conduction#

Volume conduction is the spread of electrical potentials through the tissue between the source and the electrode; it is what makes crosstalk possible.

Subcutaneous fat acts as a low-pass filter and attenuator: the thicker the tissue layer, the smaller and smoother the recorded signal.

Specific vs. quasi-specific placement#

A specific placement records predominantly one muscle; a quasi-specific placement records a muscle group whose members cannot be separated with surface electrodes.

Cram's atlas uses this distinction for every site. "Wrist flexors", "medial hamstrings" or "hip adductors" are quasi-specific by nature; the EMG Guide practice block states the type for each muscle.

Fine-wire (intramuscular) EMG#

Fine-wire EMG records from within the muscle through thin hooked wires inserted with a hypodermic needle; it is required for muscles that are deep or covered by other muscles.

Levator scapulae, rhomboids, supraspinatus, iliopsoas or tibialis posterior belong to this group — which is why they are not part of the EMG Guide catalogue.

Amplifier and specifications#

Common-mode rejection ratio (CMRR)#

The CMRR states how strongly a differential amplifier suppresses signals that are identical at both inputs, expressed in decibels; SENIAM recommends at least 95 dB.

Mains hum and ECG reach both electrodes almost equally and are therefore attenuated — the higher the CMRR, the cleaner the baseline.

Input impedance#

Input impedance is the resistance the amplifier presents to the electrodes; it must be much higher than the skin impedance so that almost no current flows and the signal is not attenuated.

SENIAM recommends at least 100 MΩ. Modern amplifiers reach gigaohm values, which makes them more tolerant of imperfect skin preparation.

Gain#

Gain is the factor by which the amplifier enlarges the microvolt-range EMG signal before it is digitised.

Typical values are 500–2000. Too high a gain clips large signals; too low a gain wastes the resolution of the analog-to-digital converter.

Bandwidth / bandpass filter#

The bandpass defines the frequency range that is passed by the amplifier; SENIAM recommends a high-pass of 10 Hz and a low-pass of about 500 Hz for surface EMG.

Most of the signal power lies between 20 and 250 Hz. A 20-Hz high-pass reduces motion artifacts at the cost of some real signal, and is common in dynamic tasks.

Sampling rate#

The sampling rate is the number of samples per second the analog signal is digitised with; it must be at least twice the highest frequency in the signal (Nyquist), i.e. ≥ 1000 Hz for a 500-Hz low-pass.

Undersampling produces aliasing — high frequencies masquerade as low ones and cannot be removed afterwards.

Resolution (bit depth)#

Resolution is the number of discrete levels the analog-to-digital converter can distinguish, given in bits; 12 bits give 4096 levels, 16 bits 65 536.

Together with gain and input range it determines the smallest voltage step that can be represented; 16 bit or more is standard today.

Noise floor#

The noise floor is the signal level an amplifier produces with no muscle activity — the electronic noise of the system itself.

Konrad recommends a raw baseline with a rectified mean of about 1–3.5 µV; spikes above 10–15 µV point to a preparation or environment problem rather than to the amplifier.

Signal-to-noise ratio (SNR)#

The signal-to-noise ratio compares the amplitude of the EMG signal with that of the noise; it decides whether small activations can be detected at all.

Thick subcutaneous fat, poor skin preparation and small muscles lower it. A muscle that reads "silent" may simply be recorded with too poor an SNR.

Signal processing#

Raw EMG#

The raw EMG is the amplified, bandpass-filtered but otherwise unprocessed signal — a zero-centred trace with positive and negative spikes.

It is the only trace that shows artifacts as they are, which is why every signal check starts with the raw signal, not with a smoothed envelope.

Rectification#

Rectification converts the raw EMG into a signal with only positive values, either by inverting the negative half (full-wave) or by cutting it off (half-wave).

It is a prerequisite for averaging, because the mean of a raw EMG is zero.

Smoothing / envelope#

Smoothing produces the envelope of the rectified EMG by averaging over a moving window; the result reflects the amplitude course of the activity.

Typical windows are 50–100 ms for fast movements and 100–250 ms for slow or static tasks. Longer windows give a calmer curve but blur timing.

RMS (root mean square)#

The RMS is the square root of the mean of the squared signal over a window; it is the recommended amplitude measure in surface EMG because it reflects the signal power.

RMS values are stated in microvolts (µV). Compared with the average rectified value it is less sensitive to amplitude cancellation and easier to relate to physiological quantities.

Average rectified value (ARV / MAV)#

The average rectified value is the mean of the rectified EMG over a window — the simpler alternative to RMS.

The two are strongly correlated; what matters is to use one measure consistently and to state which.

Integrated EMG (iEMG)#

The integrated EMG is the area under the rectified curve over a time period, in µV·s.

Unlike RMS it grows with the length of the interval, so iEMG values are only comparable for equal durations. Older literature often uses "iEMG" loosely for any smoothed amplitude.

Normalisation#

Normalisation expresses the EMG amplitude as a percentage of a reference contraction — usually the maximum voluntary isometric contraction (MVIC) — instead of in absolute microvolts.

Absolute microvolt values depend on skin, fat, electrode position and equipment and cannot be compared between muscles, people or days. Normalisation is what makes "80 % of maximum" a meaningful statement.

MVC / MVIC#

The maximum voluntary (isometric) contraction is a standardised, well-braced maximal effort against fixed resistance, used as the reference for normalisation.

It requires a defined position, verbal encouragement, several trials and adequate rest; in pain or after injury a true maximum may not be reachable and sub-maximal references are used instead.

Digital filter#

A digital filter modifies the frequency content of the sampled signal — high-pass to remove baseline drift, low-pass to smooth, notch to suppress mains hum.

Filters shift and blur signals in time; the same filter settings must be used within one comparison, and a notch filter should be a last resort because it also removes real EMG at 50/60 Hz.

ECG artifact reduction#

ECG artifact reduction removes the heartbeat that couples into EMG channels on the trunk and shoulders — by high-pass filtering, template subtraction or gating.

Erector spinae, rectus abdominis, pectoralis and the left trapezius are the muscles most affected.

Time normalisation#

Time normalisation stretches or compresses each repetition of a movement to a fixed length (e.g. 0–100 % of a gait cycle) so that repetitions can be averaged.

Ensemble averaging over many cycles then yields the typical activation curve — the standard representation in gait and movement analysis.

Analysis and parameters#

Onset / offset#

Onset and offset are the time points at which a muscle becomes active and inactive, usually defined by a threshold above the resting baseline.

Common thresholds are 2–3 standard deviations of the baseline or a fixed percentage of the peak. Timing analyses stand and fall with this definition.

Amplitude parameters#

Amplitude parameters describe how much a muscle is active — mean, peak, RMS or area of the envelope within a defined period.

They answer "more or less?", not "how strong?" — the EMG-force relation is muscle-specific and non-linear.

Frequency parameters — mean and median frequency#

The mean and median frequency describe the centre of the EMG power spectrum in hertz; both shift towards lower values as a muscle fatigues.

The median frequency (the frequency splitting the spectrum's power in half) is more robust against noise and is the standard fatigue index in sustained isometric contractions.

Power spectrum#

The power spectrum shows how the signal power is distributed over frequency, computed by a Fourier transform.

For a clean surface EMG it rises steeply from the high-pass, peaks around 50–80 Hz and approaches zero by 200–250 Hz. A sharp peak at 50 or 60 Hz reveals mains hum.

Muscle fatigue (in EMG)#

In surface EMG, fatigue shows as a decrease of median frequency together with a rise of amplitude at constant force during sustained contraction.

The shift reflects slowing conduction velocity and changing motor unit behaviour; it appears before force actually declines.

Co-contraction#

Co-contraction is the simultaneous activity of agonist and antagonist around a joint, quantified by the overlap of their EMG envelopes.

It stiffens the joint; the amount is task-dependent and increases in unstable or novel conditions.

Timing / muscle coordination#

Timing analysis compares onset, peak and offset of several muscles within a movement to describe their coordination pattern.

Typical questions: does the vastus medialis fire before the vastus lateralis? Does the gluteus maximus start before the hamstrings?

Test standardisation#

Test standardisation is the fixed definition of position, movement, load, speed, instruction and rest periods so that repeated measurements differ only by the quantity of interest.

Without it, differences between sessions cannot be attributed to the muscle. Konrad lists it as the precondition of any EMG comparison.

Baseline (tonic level)#

The baseline is the EMG level during intended rest; in a well-prepared, relaxed muscle it stays flat near zero and returns there within milliseconds after every contraction.

A raised or drifting baseline indicates either incomplete relaxation or an artifact — the distinction is one of the core skills in EMG biofeedback.

Artifacts#

Motion artifact#

A motion artifact is a low-frequency disturbance caused by relative movement between electrode, cable and skin — visible as slow baseline waves or sharp deflections.

Countermeasures: thorough skin preparation, cable fixation with tape or netting, and a 20-Hz high-pass in dynamic tasks.

Mains (power line) hum#

Mains hum is a periodic 50-Hz (Europe) or 60-Hz (Americas) interference coupled from the electrical environment into the recording.

Causes are usually poorly grounded devices nearby (treadmills, training machines) or a bad reference electrode; grounding and cable routing solve it more reliably than a notch filter.

ECG artifact#

The ECG artifact is the heartbeat appearing as regular spikes roughly once per second in EMG channels near the thorax.

It is physiological, not a fault — but it inflates amplitude values of trunk muscles and must be recognised or removed.

Baseline offset and shift#

A baseline offset is a constant displacement of the raw EMG from zero; a baseline shift is a temporary deviation that does not return to zero within a few milliseconds.

Offset is corrected before recording; shifts indicate cable movement, pressure on the electrode or muscle wobble and invalidate amplitude values in that interval.

Standards and references#

SENIAM#

SENIAM (Surface ElectroMyoGraphy for the Non-Invasive Assessment of Muscles) is the European project (1996–1999) that published consensus recommendations for sensor properties, sensor placement and signal processing in surface EMG.

Its placement recommendations cover some 30 muscle sites and are the basis of the 52 SENIAM-marked muscles on EMG Guide; the site's remaining muscles use placements from peer-reviewed sources named on each page.

ISEK#

The International Society of Electrophysiology and Kinesiology publishes reporting standards for EMG studies — which sensor, filter, sampling and normalisation details must be stated in a paper.

Following the ISEK checklist makes a measurement reproducible by others; it is a good template for a practice protocol too.

Kinesiological vs. clinical EMG#

Kinesiological EMG studies muscle activation during voluntary posture, movement and training; clinical (neurological) EMG uses needle electrodes and electrical stimulation to diagnose nerve and muscle disease.

Everything on EMG Guide belongs to the first kind. Surface EMG visualises activation patterns; it does not diagnose.