the emg signal filtering signal processing.2. filters - overview u primary function is noise...

Post on 31-Mar-2015

216 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

The EMG Signal

Filtering

Signal Processing.2

Filters - Overview

Primary function is noise attenuation If the frequency of the noise source is

sufficiently different from the frequency components of the signal waveform of interest - the noise can be removed providing a “cleaner” EMG signal

Filters - Overview

Frequency range of muscle - slow twitch motor units

(20) 70 - 125 Hz

Filters - Overview

Frequency range of muscle - slow twitch motor units

Fast twitch motor units

125 - 250 Hz

Filters - Overview

Frequency range of muscle - slow twitch motor units

Fast twitch motor units

Sources of noise that “compete” with these frequency ranges

Attenuate or make the true signal less visible and difficult to interpret– Example: 60 Hz from

120 V power lines

Filter Types

Hardware filters– Analog electronic circuit

» Amplifiers, resistors, capacitors

Software filters– “Digital” filters

» Mathematical algorithms

Butterworth Filter.vi

Frequency Components

Bandwidth– Range of frequencies from the low frequency

limit of the EMG signal to the high frequency limit = the band pass

Low frequency cut-off High frequency cut-off Roll off

– Rate at which frequency attenuation occurs

Frequency Components

Filter Types by Frequency Component

LP Filter

20 Hz 250 Hz

High frequency filter– Removes high

frequency components above a certain “cut-off”

Low pass filter (LP)– Pass = retain

Filter Types by Frequency Component

BPFilter

20 Hz 250 Hz

Low - High frequency filter– Removes frequency

components below and above certain “cut-offs”

Band pass filter (BP)Filter

Filter Types by Frequency Component

BS

20 Hz 250 Hz

Mid-range frequency filter– Removes a specific

frequency component within a range

Band stop filter (BS)– Example: 60 Hz filter

60 Hz

Roll Off

Rate at which frequency attenuation occurs

Expressed by the order– The higher the

order the more rapid the roll off

– Index of sensitivity of attenuation

2

B u tte rw o rth F ilte r.v i

2

Butterworth Filter.vi

Phase Shift

Filtering causes a change in phase = shift– A time delay frequency

component as it passes through the filter

– May cause waveform distortion especially if the the shift occurs near the cut-off frequency

If the phase shift is small it may be a tolerable error source

Shift

Phase Shift Solution

Filter Use Turn filter “On” Select type Insert cut-off(s) Run the VI

Practical Effect - Filtering

Filtering will “sharpen” the image permitting better approximation of important events– Onsets

– Offsets

– Etc.

Practical Effect - Filtering

Raw

Filtered

Signal Processing.2

Descriptive statistics– Signal spike counts– Peak amplitude (voltage - mV) detection– Averaging– Variability analysis

Root Mean Square

Descriptive Statistics

Often used as a basic means of analysis after visual inspection of the raw data– Probably more useful in quantifying “On-Off”

phenomena– May be used in conjunction with time-based

analyses: onset, duration & offset

Signal Spike Counts

More useful when muscle force levels are relatively low– The interference pattern typical of high force

levels (e.g., MVC) makes spike counts difficult

4.0

-2.0

0.0

2.0

20000 500 1000 1500

Signal Spike Counts

Spike Counting by Window Spike Counting of RawSignal - (could also be donewith rectified signal)

Peak Amplitude Has traditionally been issued as an index of

maximal muscle activity– Probably valid when electrical activity is

relatively constant– A peak amplitude that exists more as an outlier

may not be truly representative of typical or average activity

Full-Wave Rectified Signal

Averaging (Mean)

A data smoothing technique useful when high signal variability is of concern

Moving average - the mean amplitude of a full-wave rectified window (segment) of data points for:– Baseline noise (last session: “2 SD Method”)– The true EMG signal

Ensemble average - a mean of mean segments across subjects or trials

Variability Analysis.1 Reproducibility of

recording electrodes (e.g., Δ’s in skin resistance; number of motor units sampled) with repeated measures designs is problematic– Within subjects (e.g.,

over several days)

– Between subjects

Report EMG amplitude (e.g., mean amplitude or integral - next session) as a percentage of a baseline MVIC

Variability Analysis.2

Variability assessment of EMG will document reproducibility/consistency– SD: measure of dispersion about the mean

stated in units of interest (mV)– CV: describes dispersion of a group mean as a

percentage– SE: low SE argues sample mean is a good

estimate of the population mean

Root Mean Square (RMS)

One of several methods of quantifying EMG output (in mV) using– Hardware

or

– Software The “effective” value (quantity) of the EMG

signal (i.e., not an average) Measures electrical power A form of linear envelope procedure

RMS Value Reflects

Motor unit– Firing rates– Duration– Velocity of the electrical signal

Electrode configuration Instrument (amplifier characteristics)

RMS Procedure

Individual amplitudes are squared A mean of the squared amplitudes is

calculated Square root is calculated

RMS - Time Constant Selected

Hardware

RMS - Time Constant Selected

Hardware Software

RMS - Time Constant

Should be consistent with the nature of the activity being performed– Slow movement

» Use a longer time

– Fast movement» Use a shorter time

Reference Sources

Gitter, A.J., & Stolov, W.C. (1995). AAEM minimongraph #16: instrumentation and measurement in electrodiagnostic medicine-part I. Muscle & Nerve, 18, 799-811.

Reference Sources

Soderberg, G.L., & Knutson, L.M, (2000). A guide for use and interpretation of

kinesiologic electromyographic data. Physical Therapy, 80, 485-498.

Winter, D.A. (1990). Biomechanics and motor control of human movement (2nd Ed). New

York: John Wiley & Sons, Inc., 191-212.

top related