Hello everyone,
Have you ever imagined visualizing muscle forces ?? There are many ways, but electromyography aka EMG is just one way. It detects action potentials on muscles and shows signal waves as an output. As a simple example, the bigger waves, the more muscle force. In this blog, I will show you how to collect and analyze EMG data.
What does EMG look like ??
You can watch the video applying EMG on hamstring. On the video, 2 small poles detect action potentials, and 1 blue box sends the detected data to a receiver which is connected to an AD converter. EMG placement depends on muscles. For guidance, please visit SENIAM.
What does Nordic Hamstring Curl look like ??
As you can see, a subject goes down as deep as possible. Of course, while he is doing, EMG receives signals which are visible on screen.
We had 5 trials in total and collected signals on right and left semitendinosus and bicep femoris.
Procedure
1, Import data
I downloaded data from proEMG and imported it into MATLAB. The raw data looks like …
2, Detrend data
Detrending means correcting a base line to 0. Sometimes, EMG signals are not based on 0. For example, resting EMG is 0 as you imagine, but sometimes, it is not 0 for some reasons. Then, detrending correct the base line to 0. The detrended data looks like …
It seems like no difference from the raw data … Maybe the raw data is already detrended.
3, Power spectrum
Power spectrum is to see what creates the EMG signal (like a recipe). The power spectrum looks like this …
As you can see, there are some strong frequencies between 10 and 450Hz. It would suggest that we should remove all other frequencies and take only frequencies within the band (Bandpass filtering).
4, Butterworth filter
As I mentioned above, we should get rid of noise (remove frequencies outside of the band). Bandpass filtering does the job. A common filtering routine for EMG is bandpass filtering with cutoff frequencies of 10 and 500. In this time, the power spectrum suggests that range, so this is exactly the case. Filtered data looks like this …
You may see some small changes, but not huge. Maybe, it is due to a small amount of noises.
5, Rectify the data
Rectifying = Flipping sign from negative to positive. In EMG signals, negative positive amplitudes mean same, so by flipping the sign, you can see some muscular events more easily. Rectified data looks like this …
Maybe, now you can see some muscle on and off events as well as muscle force development, but we can still make it better !!
6, Moving average
Moving average is an algorithm to smooth out the data. Simple definition is averaging subsets of data in order. This process especially on EMG is known as enveloping. The enveloped data looks like this …
Now you can clearly see muscle on and off as well as force development.
My YouTube video
I am explain how to analyze the data on my YouTube video, so please check it out !!
MATLAB code and EMG data
Enjoy !!
