ELECTROMIOGRAPHY

Dr. Amit M. Joshi

Assistant Professor,
Dept. of Electronics & Communication
Malaviya National Institute of Technology
Jaipur (Rajasthan)
What is EMG?
• EMG stands for Electromyography(ELECTRIC-MUSCLE-GRAPH)
• Electromyography (EMG) is used to record the electrical activity of muscles and nerves.
• EMG signal is an important biopotentials signal which helps to sense the muscular
activity of the human body
• Two kinds of EMG signals in widespread use include surface EMG, and intramuscular
EMG.
• In intermuscular EMG a needle electrode is placed within the muscle of interest (Invasive
Technique)
• However surface electrodes is are non-invasive, and the patient need not be
anesthetized before placing the electrode. The operation is simple and painless.

Fig 1 :EMG signal

How EMG signal is generated ?
• When muscle of body is recruited for certain activity, brain sends excitation signal to
motor unit through CNS.
• Motor unit is point where neuron and muscle fiber meet.
• When motor unit is activated it produces MUAP(motor unit active potential).
• The activation is continued till required force is not generated and continuous trains are
produced .This trains are superimposed to produced EMG.

Fig 2 :motor unit

 EMG Signal Characteristic
Surface Electrodes Intramuscular Electrodes or
(sEMG) needle electrode
(iEMG)

Amplitude: 50 uV to 20 mV
Frequency: 5 Hz to 500 Hz
(Dominant)
 Electromyography(EMG) Signal

S. Pancholi and A. M. Joshi, "Improved Classification Scheme Using Fused Wavelet Packet Transform Based
Features for Intelligent Myoelectric Prostheses," in IEEE Transactions on Industrial Electronics, vol. 67, no.
10, pp. 8517-8525, Oct. 2020, doi: 10.1109/TIE.2019.2946536.
 EMG Signal Generation
Cortex
Peripheral
Nervous
System

Spinal
Cord
S. Pancholi and A. M. Joshi, "Time Derivative Moments Based Feature Extraction Approach for Recognition of
Upper Limb Motions Using EMG," in IEEE Sensors Letters, vol. 3, no. 4, pp. 1-4, April 2019, Art no. 6000804,
doi: 10.1109/LSENS.2019.2906386.
Characteristics of EMG signal

Fig 3 :Characteristics of EMG signal

• Amplitude can range between 0 to 10 millivolts (peak-to-peak) or 0 to 1.5 mill volts
(rms).
• The frequency of an EMG signal is between 0 to 500 Hz. However, the usable energy of
EMG signal is dominant between 50-150 Hz
Noise affecting to the EMG Signal
• EMG signal consists of two types of noises transducer noise and ambient noise
• Ambient Noise: Electromagnetic devices cause ambient noise for example computers,
power lines, force plates, etc. The dominant frequency is corresponding to the frequency
of the AC power supply i.e. 50Hz or 60Hz.
• Transducer Noise: Transducer noise is from the junction of the electrode and skin.
Surface electrodes convert the muscles generated current into electronic current. Two
types of noises result from this transduction noise.
1)DC Voltage Potential: This is due to the impedance difference between the electrode
sensor and skin.
2)AC Voltage Potential: It is produced by impedance fluctuations between the skin and
conductive transducer.
• The Noise in the EMG signal is reduced by technological development. The most
important development is to introduce a bipolar recording method. To suppresses,
signals that are common to input electrodes in Bipolar electrode organizations used
differential amplifier
Signal collection and preprocessing:

• Two distinct methods namely invasive and non- invasive techniques used to obtain EMG
signals .
• Intramuscular EMG electrodes are inserted into the muscle in invasive technique ,while
surface electrodes are applied to the skin surface in non-invasive technique .
• IEMG signal is having negligible crosstalk and can be use to obtain signals from deep
muscles ,however repeated use of this technique is practically infeasible .
• Unlike IEMG ,SEMG electrodes can collect electrical activity data from multiple muscle
and deliver sufficient neural information.
• sEMG signal acquisition performed in 2 stages. The first stage contains electrode-
amplifier stage that includes detection of signal and preamplification. The second stage is
signal conditioner stage that includes filters for noise removal and analog to digital
converters.
• A band-pass filter of 20–500 Hz bandwidth used and wiener filter is used as an effective
tool to suppress involuntary background spikes.

Fig 4:Signal acquisition and preprocessing -I

Data Segmentation:

• There is need of feature extraction when data is processed for feature extraction .
• It confirms the continuous disbursement of information to make feasible for
classification .
• Two types of signal segmentation :
1)overlapping
2)non overlapping
• In non overlapping, new segment is processed without any overlapping with predefine
window length
• Overlapping approach is mostly preferred because it yields high accuracy over non
overlapping approach.
• Window size and overlap length are two important parameters of windowing .
• Window size of 200ms with 100ms overlapping is widely used ,as it gives high
classification accuracy and meet the timing constraints of response time.

Here PT is Processing Time

Feature Extraction Methods:

Generally, features in analysis of the EMG signal can be divided into three main groups.
There are time domain, frequency domain, and time-frequency groups.

1)Time domain features:

• Features in time domain are usually quick and easy implemented, because these features
do not need any transformation, which calculate based on raw EMG time series
• Time domain features have been widely used in both medical and engineering
researches and practices.
• Major disadvantage of TD features is ,EMG data is non-stationary but time domain
features assume the data as a stationary signal.
Here are some TD features:

• Integrated EMG :Integrated EMG (IEMG) is normally used as an onset detection index in
EMG.

• Mean absolute value: MAV is defined as the average of EMG signal's total absolute
values. It can be expressed as:

• Root Mean Square: it relates to constant force and non-fatiguing contraction.

• Simple square integral :Simple square integral (SSI) or integral square uses energy of the
EMG signal as a feature.

• Waveform Length : The Waveform Length (WL) is intuitively the cumulative length of the
waveform over the segment. The resultant values of the WL calculation indicate a
measure of the waveform amplitude, frequency, and duration
It is the improvement of IEMG feature . It is defined as the cumulative length of waveform
over the segment.

• Variance: Variance of EMG (VAR) is used as power index

• Zero Crossing: Zero crossing (ZC) is a measure of frequency information of the EMG
signal that is defined in time domain. It is a number of times that amplitude values of the
EMG signal cross zero amplitude level(or threshold)
• Slope Sign Change: It is another method to represent frequency information of the EMG
signal.

• Auto Regressive Coefficient: AR model tries to model the signal by calculating previous
samples of the signal. In this each sample represents the linear combination of previous
samples.it is defined as:

• Willison amplitude: Willison amplitude or Wilson amplitude (WAMP) is a measure of

frequency information of the EMG signal as same as defines in ZC feature. It is a number
of times resulting from difference between the EMG signal amplitude among two
adjoining segments that exceeds a pre-defined threshold.
• Willison amplitude (WAMP) is the number of. counts for each change in the EMG
signal. amplitude that exceeds a predefined threshold. where threshold value is 10 mV.
Frequency Domain Features

• Frequency or spectral domain features are mostly used to study fatigue of the muscle
and MU recruitment analysis.
• Power spectral density (PSD) becomes a major analysis in frequency domain.
• FD features require more computation time because they require transformation but FD
features are more accurate than TD features.
Here are some FD features
• Mean frequency :MNF is an average frequency which is calculated as sum of product of
the EMG power spectrum and the frequency divided by total sum of the spectrum
intensity.

• Peak Frequency:At Peak frequency maximum power is obtained in EMG power

spectrum.
• Mean Power: MNP is an average power of the EMG power spectrum.

• Median frequency :MDF is a frequency at which the spectrum is divided into two regions
with equal amplitude.

• Frequency ratio: Frequency ratio (FR) is proposed to distinguish between contraction

and relaxation of muscle using ratio between the low frequency components and the
high frequency components of the EMG signal

• Power spectrum ratio: The PSR is defined as ratio between the energy P0 which is
nearby the maximum value of the EMG power spectrum and the energy P which is the
whole energy of the EMG power spectrum.
Time-Frequency Domain Features

• Contains both time and frequency information and shows enhancement in pattern
classification rate.
• Most common features are wavelet transform (WT) and Short Time Fourier
Transform(STFT),Wavelet Packet Transform(WPT) and B-distribution(BD).
• This are not used widely because they cannot be directly used and their computation is
complex also their dimensionality is high
 Other Bio-signals and sensing techniques:
1)Mechanomyography:
• MMG is an alternative to EMG that measures muscle vibrations (i.e. mechanical activity) using a
sensor, such as a microphone or an accelerometer
• Mechanomyography (MMG), also called vibromyography, is another technique for monitoring
muscle activation that can be used to design HMIs for prosthesis control.
• MMG tracks the vibrations emitted from a muscle during its contractions.
• MMG can be noninvasively acquired from the surface of a user’s skin through affixing
microphones, accelerometers, and piezoresistive sensors.
2) Force myography:
• Force myography (FMG) monitors a limb’s volumetric changes, which are indicative of muscle
contractions/relaxations.
• This technique has been used in a variety of applications to identify hand and wrist movements
• In comparison with the more traditionally accepted sEMG-based methods, FMG demonstrates
comparable performance with less sensitivity to environmental variables and with reduced cost
3)Electroneurography:
• Electroneurography is a physiologic test that uses EMG to objectively measure the difference
between potentials generated by the facial musculature on both sides of the face in response to a
supramaximal electrical stimulation of the facial nerve.
• Electroneurography (ENG) is a technique that acquires signals directly from the peripheral
nervous system through the intraneural implantation of the interfacing electrodes.
• This technique can provide sensory feedback and potentially restore the sensory function of the
lost limb.
• ENG-based HMIs were preliminarily introduced for neural spike detection ,their application has
recently been extended to detecting grasp gestures in prosthetic devices

4)Electroencephalography(EEG) :
• EEG signals are collected by placing electrodes (often a high density of sensors) on the user’s
scalp.
• Compared to other bio signals, such as FMG, EEG data require more pre-processing to remove
artifacts and to increase the accuracy of intention prediction.
• Despite their added computational cost, since EEG signals are acquired from the scalp and their
acquisition does not depend on the condition of the residual limb’s muscles.
Applications of EMG Signals
For Neuromuscular diseases and Disorders of motor unit :
1)The amyotrophic lateral sclerosis (ALS)
• is a one type of motor neuron disease which affects both the lower and upper motor neurons
and causes the muscles to become weaker and smaller
• Body becomes paralyzed progressively, which means that the muscles stop working.
• Besides clinical analysis, EMG signal analysis can also be used to conclusively know the
existence of ALS, based on EMG data decomposition.
• Features generally used are RMS and Standard deviation .

2) Parkinson's disease:
• brain disorder that leads to shaking, stiffness, and difficulty with walking, balance, and
coordination
• Attacking 1 percent of the elder population in this world, with one of the main symptoms of
this disease being resting tremor and It is triggered by a deficiency of dopamine
• Resting tremor occurs when the muscle is relaxed, such as when the hands are resting on the
lap.
• symptoms of Parkinson's disease include bradykinesia or low voluntary movement initiation
followed by lack of energy in doing repetitive movements, resting tremor, muscle rigidity marked
by patient’s difficulty to stand, walk, and causing them to have a stooped posture
• With the presence of biopotential technology, the muscle activity pattern of Parkinson’s disease
patients can be checked and recorded using electromyography (EMG) techniques

3) Gastrointestinal (GI) disorders:

• Gastrointestinal disorders is the term used to refer to any condition or disease that occurs within
the gastrointestinal tract.
• Gastrointestinal diseases affect the gastrointestinal (GI) tract from the mouth to the anus
• Abdominal myoelectric activity indicates gastric and intestinal motility. The motility of intestine
i.e. the propulsive activity of intestinal muscles is affected in GI disorders such as diarrhea and
constipation are the one.
• We can use surface electrodes to measure intestinal myoelectric activity in order to study
intestinal motility in case of diarrhea and constipation.
• Mel-Frequency Cepstral Coefficients (MFCCs) are extracted as features for identification and
classification of various GI disorders
• .
as a Controlling signal:

• Electromyography (EMG) signals have been widely used and applied as a control signal in
numerous man–machine interface
• It can be used for controlling Biomechatronic prosthetic limbs such as a hand , arms and
lower limbs.
• The electrical activity naturally generated by contracting muscle in a residual limb is
amplified, processed, and used to control the flow of electricity from a battery to a
motor, which operates an artificial limb
• The main challenge is how the prosthesis mimics the human arm and that depends on
several components which are Electrode Placement, Signal Conditioning, Feature
Extraction, Feature Evaluation and Classification .
Upper Limb Prosthetic Control
 EMG signal use for prosthetic
▪An EMG (electromyogram) based
anthropomorphic upper limb help the
amputated people to bring back the
functionality like real arm.
▪Developing an anthropomorphic arm
and pattern recognition of EMG signals
for actuation of hand
▪development of pattern recognition of
intention from surface EMG (sEMG) or
in the development of anthropomorphic
hand.
 Proposed Flow of AI based
myoelectric control prosthetic
 Basic Flow of Machine learning in
EMG based Prosthetic
Feature Size
Reduction

EMG signals
Feature Feature
Extraction Matrix

Classification

Windowing Training Testing

S. Pancholi and A. M. Joshi, "Portable EMG Data Acquisition Module for Upper Limb Prosthesis Application,"
in IEEE Sensors Journal, vol. 18, no. 8, pp. 3436-3443, 15 April15, 2018, doi: 10.1109/JSEN.2018.2809458.
 Flow of Myoelectric Prosthetic control
scheme

Sidharth Pancholi, Amit M Joshi, “Advanced Energy Kernel-based Feature Extraction Scheme for
Improved EMG-PR based Prosthesis Control Against Force Variation” IEEE Transaction in Cybernetics
ANY QUERIES ?