putemg_features

Dedicated EMG feature calculator for putEMG dataset https://biolab.put.poznan.pl/putemg-dataset/

Usage

putemg_features can be used as Python3 module. In order to calculate features of a given hdf5 putEMG file based on XML feature descriptor file see example below. See all_features.xml for file format and feature list along with its parameters.

import putemg_features

xml_url = 'all_features.xml'
hdf5_url = './putEMG/Data-HDF5/emg_gestures-03-repeats_long-2018-05-11-11-05-00-695.hdf5'
ft = putemg_features.features_from_xml(xml_url, hdf5_url)

It is also possible to calculate desired single feature directly on already imported putEMG record. Avaiable features and its parameters are same as in all_features.xml file. Eg.:

import putemg_features
import pandas as pd

record = pd.read_hdf('./putEMG/Data-HDF5/emg_gestures-03-repeats_long-2018-05-11-11-05-00-695.hdf5')
df1 = putemg_features.calculate_feature(record, 'ZC', window=1000, step=500, threshold=30)
df2 = putemg_features.calculate_feature(record[22.5:30.9], name='RMS', window=500, step=250)
df3 = putemg_features.calculate_feature(record[1:10][['EMG_1', 'EMG_5']], 'RMS', window=500, step=250)

Feature calculation funcions can not only be used with putEMG data, but also with any other data given in pd.Series format, eg.:

import putemg_features
import pandas as pd
import numpy as np

noise = pd.Series(np.concatenate((np.random.normal(0,1,100), 
                                  np.random.normal(0,20,100), 
                                  np.random.normal(0,5,100))))
rms = putemg_features.feature_rms(noise, 100, 100)
zc = putemg_features.feature_zc(noise, 100, 100)

Feature List

• Integral Absolute Value (IAV)
• Average Amplitude Change (AAC)
• Approximate Entropy (ApEn)
• Auto-Regressive Coefficients (AR)
• Cepstral Coefficients (CC)
• Difference Absolute Standard Deviation Value (DASDV)
• Kurtosis (Kurt)
• LOG Detector (LOG)
• Modified Mean Absolute Value Type 1 (MAV1)
• Modified Mean Absolute Value Type 2 (MAV2)
• Mean Absolute Value (MAV)
• Mean Absolute Value Slope (MAVSLP)
• Multiple Hamming Windows (MHW)
• Multiple Trapezoidal Windows (MTW)
• Myopulse Percentage Rate (MYOP)
• Root Mean Square (RMS)
• Sample Entropy (SampEn)
• Skewness (Skew)
• Slope Sign Change (SSC)
• Simple Square Integral (SSI)
• Absolute Temporal Moment (TM)
• Variance (VAR)
• V-Order (V)
• Willison Amplitude (WAMP)
• Waveform Length (WL)
• Zero Crossing (ZC)
• Mean Frequency (MNF)
• Median Frequency (MDF)
• Peak Frequency (PKF)
• Mean Power (MNP)
• Total Power (TTP)
• Frequency Ratio (FR)
• Variance of Central Frequency (VCF)
• Power Spectrum Ratio (PSR)
• Signal-to-Noise Ratio (SNR)
• Maximum-to-minimum Drop in Power Density Ratio (DPR)
• Power Spectrum Deformation (OHM)
• Maximum Amplitude (MAX)
• Signal-to-Motion Artifact Ratio (SMR)
• Box-Counting Dimension (BC)
• Power Spectral Density Fractal Dimension (PSDFD)

Dependencies

• Pandas - https://pandas.pydata.org/
• Numpy - http://www.numpy.org/
• SciPy - https://www.scipy.org/

Attributions

• PyEEG v0.4.0 - SampEn and ApEn features - GNU GPL v3 - https://github.com/forrestbao/pyeeg