Docstoc

Lec 8 - Electromyography

Document Sample
Lec 8 - Electromyography Powered By Docstoc
					   Electromyography

        ESS 5314-001
         Lecture 8
Reading: Robertson et al. Ch 8
  EMG: Techniques and Considerations

 • EMG Pathway: What are we measuring?
MUAP
       Tissue
                Skin
       Filter           Electrolyte                 Interference
                Noise                 Electrode
                          Noise                    Leads
                                        Noise
                                                           Amplifier

                                       Band Pass
                          A/D                                 Noise
                                         Filter
                    Electrodes
• 2 Basic Categories
  – Indwelling
     • Study activation of a single motor unit
     • Needle
     • Fine Wire


  – Surface
     • Study composite action of multiple motor units
         Surface Electrodes

• Single Differential
• Double Differential
• Pre-Amplified         Δ
     Raw Signal Characteristics
• Contains frequency components in the
  range of 5 to 2000 Hz

• Most signal power is between 20 – 200 Hz
  for surface EMG
  – With some frequencies up to 1000 Hz

• What are the implications for EMG sampling
  rate selection?
    Bio-Amplifier Considerations
• Input Impedance
  – Greater value means less resistance and less signal
    attenuation.

• Gain
  – Ratio of output to input voltage (100 to 10,000 Hz)
  – Set to maximize voltage range

• Frequency Response
  – Should amplify all frequency components equally and
    linearly (bandwidth 10 to 1000 Hz)

• Common Mode Rejection Ratio
  – Ability of system to eliminate signal (noise) common to
    all active terminals (10,000:1 = 80 dB)
         System Recommendations
     Differential Electrode Configuration

• Detection Surfaces
  – 2 parallel bars, each 1.0 cm (L) x 1-2 mm (W)
  – Spacing between bars is 1.0 cm apart
• Frequency Bandwidth of 20-500 Hz
• Common Mode Rejection Ratio > 80 dB
• Noise < 2 μV RMS (20-500 Hz)
• Input Impedance > 100 MΩ
                                     - C. De Luca (1997)
            Artefact and Noise

•   Cable movement artefact
•   Electrode movement artefact
•   ECG signal
•   Muscle cross talk
•   Hum (electrical ~ 60 Hz)
Skin Preparation Recommendations


•   Clean skin with alcohol (definitely)
•   Shave hair (only excess)
•   Lightly abrade skin (? - with sandpaper)
•   Use electrode electrolyte (gel, paste, cream)
    – Often built in with newer electrodes
            Electrode Placement
•   Muscle Belly
•   Avoid tendon and lateral edge of muscle
•   Align electrodes parallel with muscle fibers
•   Proximity to motor point or innervation
    zone
    – Place electrode on midline of muscle belly
      midway between tendon and nearest
      innervation zone
                     Reliability
• Inter-Day
  – 0.60 to 0.81
  – For both sub-maximal and maximal tasks
  – Not very reliable between days, thus questionable
    repeated trials data reliability
• Intra-Day
  – 0.64 to 0.91 for maximal contractions
  – 0.78 to 0.95 for sub-maximal contractions
  – Better – thus try repeated measures in same day

  ** Keep electrodes in place – do not remove between trial
                            sessions
          Magnitude of Signal

• Maximum of 5 μV range at skin
• Gain should be set to maximize range of A/D
• Convert raw Volts of A/D to mV measured at
  skin
Scaled EMG = (Unscaled EMG)(1/gain)(1000 mV/V)
           = (Unscaled EMG)(1000)(1/gain)
   Power Spectrum and Sampling

• Significant signal power between 20-200 Hz
  – Note: movement artifact lies in frequencies
    below 10 Hz


• Shannon’s Sampling Theorem
  – Sample at least 2 times highest power in signal


• Thus, sample at 2000 Hz (if you can)
         Signal Processing
                Amplitude


      Root Mean Square (Value)
Raw                              Linear Envelop (Curve)
EMG
      Rectified EMG (Curve)

                                 Integrated (Value)
  Post Processing of EMG Data
• Rectify
• Filter
  – Low pass linear envelop (~ 3 Hz cutoff)
  – Band pass filter (10 – 20 Hz, signal passes
    between)
  – Integrated EMG (IEMG)
• Normalize
• Determining onset and offset of muscle
  activation
Signal Processing
    Raw EMG
Signal Processing
Full Wave Rectified
  REMG = |raw EMG|
   Signal Processing
       Linear Envelop
Butterworth 2nd order, cutoff = 2 Hz
   Signal Processing
       Linear Envelop
Butterworth 2nd order, cutoff = 3 Hz
    Signal Processing
        Linear Envelop
Butterworth 2nd order, cutoff = 10 Hz
Filtering – Bandpass Filter
       Bandpass Filter Cutoffs 10 and 20 Hz




      Keeps EMG signal between 10 and 20 Hz
   (i.e., it passes through the filter unattenuated)
 Signal Processing
      Integrated
        T
IEMG = ∫ t │EMG(t)│dt
    Signal Processing
 Root Mean Square (RMS)
              T
RMS = (1/T · ∫ t EMG(t)2 dt) ½
                    Normalized EMG

  Norm EMG = EMG / max(EMG during norm trial)

• Normalization Trial
   –   MVIC (isometric - concentric)
   –   MVEC (eccentric)
   –   Max EMG during other strength trial
   –   Max EMG during task trial
   –   Max EMG during task trial or MVIC (whichever is greater)


• Ultimately trying to find the greatest possible EMG and relate
  the EMG during the trial to that max (% of max EMG
  possible). But this needs to be done consistently.
      Normalization Techniques

• % MVC – Isometric (%MVIC)
• % MVC – Isokinetic or Isotonic
• % Maximum functional reading
• % Average functional reading
          Signal Processing
          Temporal Variables


                    Burst      Muscle Firing
          Onset    Duration     Sequence

Raw EMG

          Offset     Burst     Co-Contraction
                   Frequency
    Signal Processing
Illustration of Temporal Values


        Onset      Offset




            Duration
   Onset and Offset Determination

• Dual Threshold Method
  – Filtered rectified EMG must pass a certain threshold
    combination before it is considered active (on) –
    likewise for offset (off)
  – Ex. 4% and 7%
     • EMG “burst” must achieve 7% normalized activity to be
       considered active (on)
     • Time of onset occurs at the point (during the burst) at which
       the normalized EMG crossed 4%.
   Dual Threshold Onset Method
             Normalized Rectified EMG
       1.0

       0.8

       0.6


       0.4

       0.2
0.07


0.04
         Important Considerations

• Process your normalization trial EMG exactly as you
  process your trial EMG.
   – Sample rate
   – Rectification
   – Filter or Integrate

• Rectify before filtering and/or integrating
• Normalize EMG trial data uniquely for each muscle tested
• Determine activation onset and offset from normalized or
  integrated data.
      EMG and Muscle Function
• Fatigue
  – Increased EMG amplitude
  – Decrease in motor unit firing rate (frequency)
  – More synchronous firing


• Coordination
  – Sequencing and co-contraction


• Force
  – Linear and non-linear relations
          EMG-Force Relation

               EMD            EMD


Muscle Force



   EMG



                     400 ms
                 References

•   De Luca (1997)
•   Nigg & Herzog (1996)
•   Danty & Norman (1987)
•   Soderberg & Cook (1984)
•   Winter (1990) – Chapter 8
                 Data Collection
• 4 muscles monitored
  –   Tibialis Anterior
  –   Gastrocnemius
  –   Rectus Femoris (or other quadriceps muscle)
  –   Biceps Femoris (or other hamstrings muscle)

• In processing, we will normalize EMG against
  MVIC and max EMG during task to see
  difference
• Specific Tasks TBA
  – 2 to 5 trials each task, best will be exported and used in
    assignment.
  – Vertical Jump, Biodex Trials (Isometric, Isokinetic)
                    EMG Matlab Assignment
•   Separate Assignment Handout with details
•   Load EMG data
•   Rectify all data and plot an example (raw vs rectified)
•   Find max amplitudes for each muscle during trial EMG data
•   Filter data for one trial EMG (this can be in excel or matlab – your choice) and
    compare results:
     –   Low pass linear envelop (cutoff 3Hz)
     –   Bandpass (cutoffs 10 and 20 Hz)
     –   IEMG (mean peak IEMG for the 3 squats and jumps)
     –   Plot these results for comparison (either on same graph or subplotted) with raw rectified.
           •   One plot per muscle. Or plot results from only one muscle

•   Normalize the rectified, filtered EMG (treat trial and norm trial data the same) for a
    Biodex and Jump trial and compare
     –   To MVIC for each muscle (or the muscle we have isometric data on).
     –   To max EMG occurring in the trial
     –   Plot for comparison

•   Compare results for different filters using subplots
     –   Low pass linear envelop
     –   Bandpass
     –   IEMG
     –   RMS EMG
                Reminders
• 3rd and final critique due next week
• Assignments 1, 2 & 4 due by next week at
  latest
• Assignment 3 due before Thanksgiving
• Assignments 4 & 5 due by last day of classes
  (Dec. 3)
• Project Presentations are Nov. 13 and 20th.
• Written Projects are due by last day of
  classes
• No final exam for this course.

				
DOCUMENT INFO