Electromyography analysis module

electromyography.analyzeEMG(rawEMGSignal, samplerate, preprocessing=True, lowpass=50, highpass=20, threshold=0.01, nseg=3, phasic_seconds=4)

This functions acts as entrypoint for the EMG Analysis.

• Input:

  • rawEMGSignal = raw signal as list
  • samplerate = samplerate of the signal
  • lowpass = lowpass cutoff in Hz
  • highpass = highpass cutoff in Hz
  • threshold for the evaluation of ZC,MYOP,WAMP,SSC
  • nseg = number of segments for MAVSLPk, MHW,MTW

• Output:

  • results dictionary

electromyography.butter_highpass(cutoff, fs, order=5)

This functions generates a higpass butter filter

Parameters:

• cutoff (float) – cutoff frequency
• cutoff – cutoff frequency
• fs (float) – samplerate of the signal
• order (int) – order of the Butter Filter

Returns:

butter highpass filter

Return type:

list

electromyography.butter_highpass_filter(data, cutoff, fs, order)

This functions apply a butter highpass filter to a signal

Parameters:

• data (list) – ECG signal
• cutoff (float) – cutoff frequency
• cutoff – cutoff frequency
• fs (float) – samplerate of the signal
• order (int) – order of the Butter Filter

Returns:

highpass filtered ECG signal

Return type:

list

electromyography.butter_lowpass(cutoff, fs, order=5)

This functions generates a lowpass butter filter

Parameters:

• cutoff (float) – cutoff frequency
• cutoff – cutoff frequency
• fs (float) – samplerate of the signal
• order (int) – order of the Butter Filter

Returns:

butter lowpass filter

Return type:

list

electromyography.butter_lowpass_filter(data, cutoff, fs, order)

This functions apply a butter lowpass filter to a signal

Parameters:

• data (list) – ECG signal
• cutoff (float) – cutoff frequency
• cutoff – cutoff frequency
• fs (float) – samplerate of the signal
• order (int) – order of the Butter Filter

Returns:

lowpass filtered ECG signal

Return type:

list

electromyography.getAAC(rawEMGSignal)

Get the Average amplitude change.:

AAC = 1/N * sum(|x(i+1) - xi|) for i = 1 --> N-1

• Input:

  • raw EMG Signal as list

• Output:

  • Average amplitude change of the signal

Parameters:rawEMGSignal (list) – the raw EMG signal Returns:Average Amplitude Change of the signal Return type:float

electromyography.getAFB(rawEMGSignal, samplerate, windowSize=32)

Get the amplitude at first Burst.

Reference: Du, S., & Vuskovic, M. (2004, November). Temporal vs. spectral approach to feature extraction from prehensile EMG signals. In Information Reuse and Integration, 2004. IRI 2004. Proceedings of the 2004 IEEE International Conference on (pp. 344-350). IEEE.

• Input:

  • rawEMGSignal as list
  • samplerate of the signal in Hz (sample / s)
  • windowSize = window size in ms

• Output:

  • amplitude at first burst

Parameters:

• rawEMGSignal (list) – the raw EMG signal
• samplerate (int) – samplerate of the signal int Hz
• windowSize – window size in ms to use for the analysis

Returns:

Amplitute ad first Burst

Return type:

float

electromyography.getDASDV(rawEMGSignal)

Get the standard deviation value of the the wavelength.:

DASDV = sqrt( (1 / (N-1)) * sum((x[i+1] - x[i])**2 ) for i = 1 --> N - 1    

• Input:

  • raw EMG Signal

• Output:

  • DASDV

Parameters:rawEMGSignal (list) – the raw EMG signal Returns:standard deviation value of the the wavelength Return type:float

electromyography.getFR(rawEMGPowerSpectrum, frequencies, llc=30, ulc=250, lhc=250, uhc=500)

This functions evaluate the frequency ratio of the power spectrum.

Cut-off value can be decidec experimentally or from the MNF Feature See: Oskoei, M.A., Hu, H. (2006). GA-based feature subset selection for myoelectric classification.

• Input:

  • raw EMG power spectrum as list,
  • frequencies as list,
  • llc = lower low cutoff
  • ulc = upper low cutoff
  • lhc = lower high cutoff
  • uhc = upper high cutoff

• Output:

  • Frequency Ratio

Parameters:

• rawEMGPowerSpectrum (list) – power spectrum of the EMG signal
• frequencies (list) – frequencies of the PSD
• llc (float) – lower cutoff frequency for the low frequency components
• ulc (float) – upper cutoff frequency for the low frequency components
• lhc (float) – lower cutoff frequency for the high frequency components
• uhc (float) – upper cutoff frequency for the high frequency components

Returns:

frequencies ratio of the EMG power spectrum

Return type:

float

electromyography.getHIST(rawEMGSignal, nseg=9, threshold=50)

Histograms is an extension version of ZC and WAMP features.

• Input:

  • raw EMG Signal as list
  • nseg = number of segment to analyze
  • threshold = threshold to use to avoid DC fluctuations

• Output:

  • get zc/wamp for each segment

Parameters:

• rawEMGSignal (list) – the raw EMG signal
• nseg (int) – number of segments to analyze
• threshold (int) – value to sum / substract to the zero when evaluating the crossing.

Returns:

Willison amplitude

Return type:

float

electromyography.getIEMG(rawEMGSignal)

This function compute the sum of absolute values of EMG signal Amplitude.:

IEMG = sum(|xi|) for i = 1 --> N

• Input:

  • raw EMG Signal as list

• Output:

  • integrated EMG

Parameters:rawEMGSignal (list) – the raw EMG signal Returns:the IEMG of the EMG Signal Return type:float

electromyography.getLOG(rawEMGSignal)

LOG is a feature that provides an estimate of the muscle contraction force.:

LOG = e^((1/N) * sum(|xi|)) for x i = 1 --> N

• Input:

  • raw EMG Signal

• Output = * LOG

Parameters:rawEMGSignal (list) – the raw EMG signal Returns:LOG feature of the EMG Signal Return type:float

electromyography.getMAV(rawEMGSignal)

Thif functions compute the average of EMG signal Amplitude.:

MAV = 1/N * sum(|xi|) for i = 1 --> N

• Input:

  • raw EMG Signal as list

• Output:

  • Mean Absolute Value

Parameters:rawEMGSignal (list) – the raw EMG signal Returns:the MAV of the EMG Signal Return type:float

electromyography.getMAV1(rawEMGSignal)

This functoin evaluate Average of EMG signal Amplitude, using the modified version n°.1.:

IEMG = 1/N * sum(wi|xi|) for i = 1 --> N
wi = {
      1 if 0.25N <= i <= 0.75N,
      0.5 otherwise
      }

• Input:

  • raw EMG Signal as list

• Output:

  • Mean Absolute Value

Parameters:rawEMGSignal (list) – the raw EMG signal Returns:the MAV (modified version n. 1) of the EMG Signal Return type:float

electromyography.getMAV2(rawEMGSignal)

This functoin evaluate Average of EMG signal Amplitude, using the modified version n°.2.:

IEMG = 1/N * sum(wi|xi|) for i = 1 --> N
wi = {
      1 if 0.25N <= i <= 0.75N,
      4i/N if i < 0.25N
      4(i-N)/N otherwise
      }

• Input:

  • raw EMG Signal as list

• Output:

  • Mean Absolute Value

Parameters:rawEMGSignal (list) – the raw EMG signal Returns:the MAV (modified version n. 2) of the EMG Signal Return type:float

electromyography.getMAVSLPk(rawEMGSignal, nseg)

Mean Absolute value slope is a modified versions of MAV feature.

The MAVs of adiacent segments are determinated.

MAVSLPk = MAV[k+1] - MAV[k]; k = 1,..,k+1

• Input:

  • raw EMG signal as list
  • nseg = number of segments to evaluate

• Output:

  • list of MAVs

Parameters:

• rawEMGSignal (list) – the raw EMG signal
• nseg (int) – number of segments to evaluate

Returns:

Mean absolute slope value

Return type:

float

electromyography.getMDF(rawEMGPowerSpectrum, frequencies)

Obtain the Median Frequency of the PSD.

MDF is a frequency at which the spectrum is divided into two regions with equal amplitude, in other words, MDF is half of TTP feature

• Input:

  • raw EMG Power Spectrum
  • frequencies

• Output:

  • Median Frequency (Hz)

Parameters:

• rawEMGPowerSpectrum (list) – power spectrum of the EMG signal
• frequencies (list) – frequencies of the PSD

Returns:

median frequency of the EMG power spectrum

Return type:

float

electromyography.getMNF(rawEMGPowerSpectrum, frequencies)

Obtain the mean frequency of the EMG signal, evaluated as the sum of product of the EMG power spectrum and the frequency divided by total sum of the spectrum intensity:

MNF = sum(fPj) / sum(Pj) for j = 1 -> M 
M = length of the frequency bin
Pj = power at freqeuncy bin j
fJ = frequency of the spectrum at frequency bin j

• Input:

  • rawEMGPowerSpectrum: PSD as list
  • frequencies: frequencies of the PSD spectrum as list

• Output:

  • Mean Frequency of the PSD

Parameters:

• rawEMGPowerSpectrum (list) – power spectrum of the EMG signal
• frequencies (list) – frequencies of the PSD

Returns:

mean frequency of the EMG power spectrum

Return type:

float

electromyography.getMNP(rawEMGPowerSpectrum)

This functions evaluate the mean power of the spectrum.:

Mean Power = sum(Pj) / M, j = 1 --> M, M = len of the spectrum

• Input:

  • EMG power spectrum

• Output:

  • mean power

Parameters:

• rawEMGPowerSpectrum (list) – power spectrum of the EMG signal
• frequencies (list) – frequencies of the PSD

Returns:

mean power of the EMG power spectrum

Return type:

float

electromyography.getMYOP(rawEMGSignal, threshold)

The myopulse percentage rate (MYOP) is an average value of myopulse output. It is defined as one absolute value of the EMG signal exceed a pre-defined thershold value.

MYOP = (1/N) * sum(|f(xi)|) for i = 1 --> N
f(x) = {
        1 if x >= threshold
        0 otherwise
}

• Input:

  • rawEMGSignal = EMG signal as list
  • threshold = threshold to avoid fluctuations caused by noise and low voltage fluctuations

• Output:

  • Myopulse percentage rate

Parameters:

• rawEMGSignal (list) – the raw EMG signal
• threshold (int) – value to sum / substract to the zero when evaluating the crossing.

Returns:

Myopulse percentage rate of the signal

Return type:

float

electromyography.getPSR(rawEMGPowerSpectrum, frequencies, n=20, fmin=10, fmax=500)

This function computes the Power Spectrum Ratio of the signal, 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

• Input:

  • EMG power spectrum
  • frequencies as list
  • n = range around f0 to evaluate P0
  • fmin = min frequency
  • fmax = max frequency

Parameters:

• rawEMGPowerSpectrum (list) – power spectrum of the EMG signal
• frequencies (list) – frequencies of the PSD
• n (int) – range of frequencies around f0 to evaluate
• fmin (int) – min frequency to evaluate
• fmax (int) – lmaximum frequency to evaluate

Returns:

Power spectrum ratio of the EMG power spectrum

Return type:

float

electromyography.getPeakFrequency(rawEMGPowerSpectrum, frequencies)

Obtain the frequency at which the maximum peak occur

• Input:

  • raw EMG Power Spectrum as list
  • frequencies as list

• Output:

  • frequency in Hz

Parameters:

• rawEMGPowerSpectrum (list) – power spectrum of the EMG signal
• frequencies (list) – frequencies of the PSD

Returns:

peakfrequency of the EMG Power spectrum

Return type:

float

electromyography.getRMS(rawEMGSignal)

Get the root mean square of a signal.:

RMS = (sqrt( (1 / N) * sum(xi**2))) for i = 1 --> N

• Input:

  • raw EMG Signal as list

• Output:

  • Root mean square of the signal

Parameters:rawEMGSignal (list) – the raw EMG signal Returns:Root mean square of the EMG signal Return type:float

electromyography.getSM(rawEMGPowerSpectrum, frequencies, order)

Get the spectral moment of a spectrum:

SM = sum(fj*(Pj**order)), j = 1 --> M

• Input:

  • raw EMG Power Spectrum
  • frequencies as list
  • order (int)

• Output:

  • SM of order = order

Parameters:

• rawEMGPowerSpectrum (list) – power spectrum of the EMG signal
• frequencies (list) – frequencies of the PSD
• order (int) – order to the moment

Returns:

Spectral moment of order X of the EMG power spectrum

Return type:

float

electromyography.getSSC(rawEMGSignal, threshold)

Number of times the slope of the EMG signal changes sign.:

SSC = sum(f( (x[i] - x[i-1]) X (x[i] - x[i+1]))) for i = 2 --> n-1

f(x){
    1 if x >= threshold
    0 otherwise
}

• Input:

  • raw EMG Signal

• Output:

  • number of Slope Changes

Parameters:

• rawEMGSignal (list) – the raw EMG signal
• threshold (int) – value to sum / substract to the zero when evaluating the crossing.

Returns:

Number of slope’s sign changes

Return type:

int

electromyography.getSSI(rawEMGSignal)

This function compute the summation of square values of the EMG signal.:

SSI = sum(xi**2) for i = 1 --> N

• Input:

  • raw EMG Signal as list

• Output:

  • Simple Square Integral

Parameters:rawEMGSignal (list) – the raw EMG signal Returns:SSI of the signal Return type:float

electromyography.getTM(rawEMGSignal, order)

This function compute the Temporal Moment of order X of the EMG signal.:

TM = (1 / N * sum(xi**order) for i = 1 --> N

• Input:

  • raw EMG Signal as list

• Output:

  • TM of order = order

Parameters:

• rawEMGSignal (list) – the raw EMG signal
• order (int) – order the the TM function

Returns:

Temporal Moment of order X of the EMG signal

Return type:

float

electromyography.getTTP(rawEMGPowerSpectrum)

This functions evaluate the aggregate of the EMG power spectrum (aka Zero Spectral Moment)

• Input:

  • raw EMG Power Spectrum

• Output:

  • Total Power

Parameters:

• rawEMGPowerSpectrum (list) – power spectrum of the EMG signal
• frequencies (list) – frequencies of the PSD

Returns:

total power of the EMG power spectrum

Return type:

float

electromyography.getVAR(rawEMGSignal)

Summation of average square values of the deviation of a variable.:

VAR = (1 / (N - 1)) * sum(xi**2) for i = 1 --> N

• Input:

  • raw EMG Signal as list

• Output:

  • Summation of the average square values

Parameters:rawEMGSignal (list) – the raw EMG signal Returns:the VAR of the EMG Signal Return type:float

electromyography.getVCF(SM0, SM1, SM2)

This function evaluate the variance of the central freuency of the PSD.:

VCF = (1 / SM0)*sum(Pj*(fj - fc)**2),j = 1 --> M, = SM2 / SM0 - (SM1 /SM0) **2

• Input:

  • SM0: spectral moment of order 0
  • SM1: spectral moment of order 1
  • SM2: spectral moment of order 0

• Output:

  • Variance of Central frequency of the Power spectrum

Parameters:

• SM0 (float) – Spectral moment of order 0
• SM1 (float) – Spectral moment of order 1
• SM2 (float) – Spectral moment of order 2

Returns:

Variance of central frequency

Return type:

float

electromyography.getWAMP(rawEMGSignal, threshold)

Wilson or Willison amplitude is a measure of frequency information. It is a number of time resulting from difference between the EMG signal of two adjoining segments, that exceed a threshold.:

WAMP = sum( f(|x[i] - x[i+1]|)) for n = 1 --> n-1
f(x){
    1 if x >= threshold
    0 otherwise
}

• Input:

  • rawEMGSignal = EMG signal as list
  • threshold = threshold to avoid fluctuations caused by noise and low voltage fluctuations

• Output:

  • Wilson Amplitude value

Parameters:

• rawEMGSignal (list) – the raw EMG signal
• threshold (int) – value to sum / substract to the zero when evaluating the crossing.

Returns:

Willison amplitude

Return type:

float

electromyography.getWL(rawEMGSignal)

Get the waveform length of the signal, a measure of complexity of the EMG Signal.:

WL = sum(|x(i+1) - xi|) for i = 1 --> N-1

• Input:

  • raw EMG Signal as list

• Output:

  • wavelength of the signal

Parameters:rawEMGSignal (list) – the raw EMG signal Returns:Waveform length of the signal Return type:float

electromyography.getZC(rawEMGSignal, threshold)

How many times does the signal crosses the 0 (+-threshold).:

ZC = sum([sgn(x[i] X x[i+1]) intersecated |x[i] - x[i+1]| >= threshold]) for i = 1 --> N - 1
sign(x) = {
            1, if x >= threshold
            0, otherwise
        }

• Input:

  • rawEMGSignal = EMG signal as list
  • threshold = threshold to use in order to avoid fluctuations caused by noise and low voltage fluctuations

• Output:

  • ZC index

Parameters:

• rawEMGSignal (list) – the raw EMG signal
• threshold (int) – value to sum / substract to the zero when evaluating the crossing.

Returns:

Number of times the signal crosses the 0 (+- threshold)

Return type:

float

electromyography.phasicFilter(rawEMGSignal, samplerate, seconds=4)

Apply a phasic filter to the signal, with +-4 seconds from each sample

• Input:

  • rawEMGSignal = emg signal as list
  • samplerate = samplerate of the signal

• Output:

  • phasic filtered signal

Parameters:

• rawEMGSignal (list) – the raw EMG signal
• samplerate (int) – samplerate of the signal in Hz

Returns:

the phasic filtered signal

Return type:

list