File Formats
Input data format
EMGFlow accepts data in plaintext .CSV file format. Files should have the following format:
- Row 1 - Column headers
- Col 1 - Labelled Time, and contains the timestamps of sampled data
- Col 2:n - Assumed to be sEMG or related signal data.
As an example, here is the first few rows of some sample data included in the package with added NaN values:
| Time | EMG_zyg | EMG_cor |
|---|---|---|
| 0.0005 | -0.145569 | -0.097046 |
| 0.0010 | -0.129089 | -0.100708 |
| 0.0015 | -0.118713 | -0.101929 |
| 0.0020 | NaN | NaN |
| 0.0025 | -0.104370 | -0.097656 |
| ... | ... | ... |
For proper calculations, ensure the 'Time' column has an equal difference between each sequential row.
EMGFlow uses pd.read_csv() from the Pandas package to read in files. Continuing with the documentation for Pandas, the following values are interpreted as NaN:
" ", "#N/A", "#N/A N/A", "#NA", "-1.#IND", "-1.#QNAN", "-NaN", "-nan", "1.#IND", "1.#QNAN", "<NA>", "N/A", "NA", "NULL", "NaN", "None", "n/a", "nan", "null"
Columns containing EMG data should not be named "Time", or begin with "mask_", as some functions use these as indicators.
Output Data Format
EMGFlow outputs a single plaintext .CSV file format containing features extracted from all processed input files. The output file has the following format:
| File_path | EMG_Zyg_Min | ... | EMG_Zyg_SpecPmissing | EMG_Cor_Min | ... | EMG_Cor_SpecPmissing |
|---|---|---|---|---|---|---|
| 01\sample_data_01.csv | 0.0027489562796772 | ... | 0.00525 | 0.0035784079507369 | ... | 0.0058 |
| 01\sample_data_02.csv | 0.0049906242832924 | ... | 0.00085 | 0.0040041486806442 | ... | 0.00035 |
| 02\sample_data_03.csv | 0.0020494782305965 | ... | 0.0168 | 0.0021287377144706 | ... | 0.01805 |
| 02\sample_data_04.csv | 0.0024170458470961 | ... | 0.0021 | 0.0021508113881666 | ... | 0.001 |
Within the output feature file, each muscle's features are blocked together. E.g., muscle 1 spans columns 2:36, muscle 2 spans 37:71 etc. The number of columns is dependent on the number of recorded muscles contained in the input data.
Column name definitions:
| Column name | Feature | Type |
|---|---|---|
| EMG_Min | Minimum voltage | Time-series |
| EMG_Max | Maximum voltage | Time-series |
| EMG_Mean | Mean voltage | Time-series |
| EMG_SD | Standard deviation of voltage | Time-series |
| EMG_Skew | Skew of voltage | Time-series |
| EMG_Kurtosis | Kurtosis of voltage | Time-series |
| EMG_IEMG | Integrated EMG of voltage | Time-series |
| EMG_MAV | Mean absolute value of voltage | Time-series |
| EMG_MMAV1 | Modified mean absolute value 1 of voltage | Time-series |
| EMG_MMAV2 | Modified mean absolute value 2 of voltage | Time-series |
| EMG_SSI | Simple square integral of voltage | Time-series |
| EMG_VAR | Variance of voltage | Time-series |
| EMG_VOrder | V-order of voltage | Time-series |
| EMG_RMS | Root mean square of voltage | Time-series |
| EMG_WL | Waveform length of voltage | Time-series |
| EMG_LOG | Log-detector of voltage | Time-series |
| EMG_MFL | Maximum fractal length of voltage | Time-series |
| EMG_AP | Average power of voltage | Time-series |
| EMG_Timeseries_Pmissing | Percentage of missing data | Time-series |
| EMG_Max_Freq | Maximum frequency | Spectral |
| EMG_MDF | Median frequency | Spectral |
| EMG_MNF | Mean frequency | Spectral |
| EMG_Twitch_Ratio | Twitch ratio of frequency | Spectral |
| EMG_Twitch_Index | Twitch index of frequency | Spectral |
| EMG_Twitch_Slope | Twitch slope of frequency | Spectral |
| EMG_SC | Spectral centroid of frequency | Spectral |
| EMG_SFlt | Spectral flatness of frequency | Spectral |
| EMG_SFlx | Spectral flux of frequency | Spectral |
| EMG_SS | Spectral spread of frequency | Spectral |
| EMG_SDec | Spectral decrease of frequency | Spectral |
| EMG_SE | Spectral entropy of frequency | Spectral |
| EMG_SR | Spectral rolloff of frequency | Spectral |
| EMG_SBw | Spectral bandwidth | Spectral |
| EMG_Spec_PMissing | Percentage of missing data | Spectral |
Sample Data Files
The sample data used in EMGFlow is taken from PeakAffectDS.
| PeakAffectDS Name | EMGFlow Name | Timeframe |
|---|---|---|
| 01-03-01.csv | sample_data_01.csv | 50.0005s : 60.0000s |
| 01-04-01.csv | sample_data_02.csv | 130.0005s : 140.0000s |
| 02-06-02.csv | sample_data_03.csv | 10.0005s : 20.0000s |
| 02-07-02.csv | sample_data_04.csv | 15.0005s : 25.0000s |
File Manipulations
The sample data files were manipulated in the following ways:
sample_data_01.csv: Added chunk of NaN values.sample_data_02.csv: Added scattered individual NaN values.sample_data_03.csv: Added several large chunks of NaN values creating "data islands"sample_data_04.csv: Injected a small high-intensity bandlimited noise pulse.