I'm an exercise physiology PhD student and am in the process of analysing
the data from my first study. One of the techniques I used was EMG, which
although has not been too tricky to use, is proving very troublesome to
analyse, so any advice or help would be much appreciated!
My study got cyclists to ride 4km in a time trial on a cycle ergometer
whilst connected to the EMG machine (although the EMG of the muscle is
not the major research interest in this study). The muscle connected to
the EMG was the VL, and the site was prepared by first shaving, then using
sandpaper and an alcohol wipe. Electrodes were attached and a sports
banadage was used to keep movement of wires etc. to a minimum.
Prior to the time trial cyclists completed an isometric MVC of the leg so
that I would be able to normalise EMG to an MVC in the analysis.
During the time trials EMG was collected at 2000Hz in a raw online form.
The data I want to extract is fairly basic - I only want 4 data points for
the EMG across the time trial, that is an EMG score (or an idea of
muscular activation, or central down-regulation) for every kilometre the
cyclists ride. From this I just want to plot a basic 4 point graph for EMG
against power output.
What I have done up until now is take the peak amplitude at each kilometre
for 4 pedal strokes (i.e for every kilometre take the peak amplitude at 4
sequential revolutions and then take an average of the 4) and then
normalise this to their MVC score to get a percentage. This has left me
with an activation score for every kilometre of the time trial.
This is exactly the data I want, however I am very concerned that I can't
find any previous authors who have used a similar method. Most authors
seem to use Butterworth filters and high/low pass filters.
Is what I have done acceptable? Reliable? Reporting an inference of
central down-regualtion if plotted alongside power output?
Or do I need to use filters on the data? The signal received by the EMG
has been brilliant and there appears to be no 'noise' when looking at the
data.
It sounds as though what I have done is a crude hand-cranked linear
envelope, but I may be completely wrong in this.
If anyone could give me some advice (bearing in mind that I am a complete
novice with EMG so it will need to be very much in layman's terms), or
recommend some references that have employed a similar technique to me, or
a technique I may be able to use I would be really really grateful.
Thankyou!
Lex Mauger (University of Exeter, UK)
the data from my first study. One of the techniques I used was EMG, which
although has not been too tricky to use, is proving very troublesome to
analyse, so any advice or help would be much appreciated!
My study got cyclists to ride 4km in a time trial on a cycle ergometer
whilst connected to the EMG machine (although the EMG of the muscle is
not the major research interest in this study). The muscle connected to
the EMG was the VL, and the site was prepared by first shaving, then using
sandpaper and an alcohol wipe. Electrodes were attached and a sports
banadage was used to keep movement of wires etc. to a minimum.
Prior to the time trial cyclists completed an isometric MVC of the leg so
that I would be able to normalise EMG to an MVC in the analysis.
During the time trials EMG was collected at 2000Hz in a raw online form.
The data I want to extract is fairly basic - I only want 4 data points for
the EMG across the time trial, that is an EMG score (or an idea of
muscular activation, or central down-regulation) for every kilometre the
cyclists ride. From this I just want to plot a basic 4 point graph for EMG
against power output.
What I have done up until now is take the peak amplitude at each kilometre
for 4 pedal strokes (i.e for every kilometre take the peak amplitude at 4
sequential revolutions and then take an average of the 4) and then
normalise this to their MVC score to get a percentage. This has left me
with an activation score for every kilometre of the time trial.
This is exactly the data I want, however I am very concerned that I can't
find any previous authors who have used a similar method. Most authors
seem to use Butterworth filters and high/low pass filters.
Is what I have done acceptable? Reliable? Reporting an inference of
central down-regualtion if plotted alongside power output?
Or do I need to use filters on the data? The signal received by the EMG
has been brilliant and there appears to be no 'noise' when looking at the
data.
It sounds as though what I have done is a crude hand-cranked linear
envelope, but I may be completely wrong in this.
If anyone could give me some advice (bearing in mind that I am a complete
novice with EMG so it will need to be very much in layman's terms), or
recommend some references that have employed a similar technique to me, or
a technique I may be able to use I would be really really grateful.
Thankyou!
Lex Mauger (University of Exeter, UK)