1) Should the signal be processed by bandpass filter (20 to 450Hz) prior to the rectification

ans: yes you should band pass it to remove high frequency noises.

2)Why did they choose 2.5Hz? Did they do it to smoothen the signals only? Why didn't they use 10Hz or above?

the low pass signal is to identify the envelope  of the emg,lets say to smoothen it up.

the cutoff frequency is not fixed, i use nearly 6-12hz for upper limb muscles, but fc generally related to length of the muscle twitch.

3) rms value, /moving window: 

better rectify and ringd the rms feature: its just the energy contained in the signal... (so it should not make any difference),

the moving window, is also a kind of filtering of emg.

Hi Aamani,

Thank you for your suggestions. Someone may argue that since it is a passive task, it would expect to have relative higher proportion of low frequency signals (i.e. may be due to slow twitch muscle activation). 

1) Will the choice of low pass filter in the signal smoothing process improve the chance of detecting the low frequency signals and get rid of the higher frequency noise? I think the choice of low pass filter doesn't matter because the filter only reduce the fluctuation of the observed signals, the low pass filtered EMG signals still contain signal frequency ranging from 20 to 450Hz. Am I right?  

2) Some scientists suggest that the purpose of using the low pass filter is to tune to match the natural frequency of muscle responses. In other words, the low pass filter is to produce a phase lag to mimic the electromechanical delay seen in vivo. The filtered signal will be more closely related to the force generated by the muscle of interest. However, if I don't know the natural frequency of muscle response and choose to use RMS rather than rectification, can I evade the use of low pass cutoff frequency?

3) if I simply calculate the RMS EMG over the entire period of passive poking instead of using the moving RMS window + calculating the average RMS over the passive poking period, the resulting RMS values from both methods should be more or less the same, right? It is because the moving window just smooths out the curve. The contained energy should be more or less the same.  

Sorry for my ignorance. 

Dear Mr. Wong

Low pass filter is a type of filter that allows lower frequency content of the signal pass and then attenuated the higher frequencies. High pass filter is doing the reverse (Lower frequencies were attenuated from the signal scale). Cut of frequency is a limit for identifying the frequency content that should be gone or stayed for further analysis. It was illustrated that the muscle signal frequency content is between 10 to 500 Hz. Also, 20 to 450 Hz was demonstrated in recent studies too (Deluca).These statements show that the frequency content lower than 10 Hz or 20 Hz and higher than 450 Hz or 500 Hz is classified as noise which contaminated the natural signals. Because of the High frequency noises (result from electrical plugs and …) and low frequency noises (result from motion artifacts or…) which contaminated the signal, researchers were utilizing both low and high pass filter for extracting the targeted signal. Then after applying the filters the amplitude analysis was initiated (different type of analysis exists like frequency one, neural network analysis and…). Linear envelope and RMS are two type of amplitude analysis of EMG signals. In Linear envelope analysis, signal rectification was performed and then the low pass filter was applied (this filter were different from previous ones that I stated before). This approach were called linear envelope. Cut of frequency quota was obtained by reading previous research or by applying residual analysis on the signal. But the RMS technique is different. In this approach the window size were like the low pass filter. Although some form of low pass filter was applied in this technique too. The result could be the same. But it is important to know that if you want to compare the result of RMS with linear envelope, applying the right window size is important.

Help this can help you.

Alighanbari.

hi, i agree with all the points mentioned above.

i addition to that linear envelope , integrated emg is also another type of pre-processing technique. please refer to that also.

thanks ,

aamani