EMG processing for forward dynamic simulation during impact events

[Pavlos Silvestros]

Dear All,

I am using EMG data collected bilaterally from the neck muscles (sternocleidomastoid and upper trapezius) during a rugby scrum engagement as controllers for a forward dynamics simulation.The impact is causing some high amplitude spikes even after a bandpass of 20-200 Hz. I have tried a few ways to smooth the data to input into the simulation (moving average, secondary lowpass [5-10 Hz] to create linear envelope), however I am not sure how to best treat the data so that I don't get "jumpy" movements from muscle activations during the simulation.

The time period I am interested in is <200-400 ms during which the amplitude alternates quickly by. Is there a preferred method of processing the data to be smooth enough for input as a controller into a forward dynamics simulation but still retain the physiological pattern or is it just trial and error?

Your help and directions would be greatly appreciated!

Best regards,
Pavlos

[Ton van den Bogert]

Pavlos,

You could try a different bandpass filter for the raw EMG. 30-500 Hz or even 50-500 Hz might remove the movement artifact from the impact. In the past, I had good results with a 30-1000 Hz bandpass filter for impact movements (http://www.me.utexas.edu/~neptune/Paper ... 282%29.pdf). When I presented this at the ISEK conference in 1998, an EMG expert was critical and said that to much low frequency EMG was removed. But you have to do what is necessary to remove the movement artifacts.

By the way, you would probably get better responses if you post this on the Biomch-L forum (http://biomch-l.isbweb.org) instead of the Opensim forum.

Ton van den Bogert

[Pavlos Silvestros]

Dear Professor,

Thank you for your reply and help, I have tried a few methods and found one that I am now happy with.

Best regards,
Pavlos