04-processing4-processing04-processing04-processing.ppt

Processing EMG David DeLion UNLV Biomechanics Lab Why do we process EMG? Raw EMG offers us valuable information in a practically useless form Raw EMG signals cannot be quantitatively compared between subjects If electrodes are moved raw EMG signals cannot be quantitatively compared for the same subject Types of Signal Processing Raw Half-wave rectified Full-wave rectified Filtering Averaging Smoothing Integration Root-mean Square Frequency spectrum Fatigue analysis Number of Zero-crossings Amplitude Probability Distribution Function Wavelet Removing Bias Low amplitude voltage offset present in hardware Can be AC or DC Calculate the mean of all the data Subtract mean from each data point Raw EMG Unprocessed signal -Amplitude of 0-6 mV -Frequency of 10-500 Hz Peak-to-Peak -Measured in mV -Represents the amount of muscle energy measured Onset times can be determined Analysis is mostly qualitative Rectification Only positive values are analyzed -Mean would be zero Half-wave rectification - all negative data is discarded, positive data is kept. Full-wave rectification- the absolute value of each data point is used Full-wave is preferred Filtering Notch filter -Band reject filter; usually very narrow -For EMG normally set from 59-61 Hz -Used to remove 60 Hz electrical noise -Also removes real data! -Too much noise will overwhelm the filter Filtering Band Pass filter -allows specified frequencies to pass -low end cutoff removes electrical noise associated with wire sway and biological artifacts -high end cutoff eliminates tissue noise at the electrode site -often set between 20-300 Hz Filtering There are no perfect filters! Face muscles can emit frequencies up to 500 Hz Heart rate artifact can be eliminated with low end cutoffs of 100 Hz Filters which include 60 Hz include the noise from equipment Averaging Average EMG can be used to quantify muscle activity over time Meas