Announcement

Collapse
No announcement yet.

Representing muscle activity: Integrated EMG (iEMG) vs Root Mean Squared (RMS)

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Representing muscle activity: Integrated EMG (iEMG) vs Root Mean Squared (RMS)

    Hello all,

    I am currently designing a study to investigate the effect of a mechanics training program on the energetics (VO2) and muscle activation during gait. I know there are different ways to process and represent EMG data including representing muscle activity as the average of Root Means Square ( RMS) for different phases of gait cycle (stance and swing) or alternatively representing the muscle activity as integrated EMG (iEMG) for each phase.

    My understanding is that iEMG is the area under the curve and RMS is the average of the squared values (using…say….a 40 ms moving window). My questions are 1) for comparing pre to post muscle activation which is the best way to represent the data? 2) What are some examples of when to use each method?

    Also, what would be the best way to normalize the amplitude of each muscle signal? MVC?

    Thank you for your time and thoughts,

    Ricardo Sanchez

  • #2
    I'd recommend reading Gait Analysis: Theory and Application by Rebecca L. Craik PhD PT, Carol S. Oatis - It's an excellent collection of explanations of all the different aspects of Gait Analysis written by many different contributors, and as a result it includes a lot of very good descriptions of EMG measurement and processing methods by highly respected authors. It was published in 1995 so it's a classical collection, but it can be found on Amazon and other book sites.

    You can normalize muscle contraction measurements by MVC but my opinion is that while it is easy to perform the testing, performing the MVC collection and analysis in a way that allows you to verify the data accuracy can be significantly affected by the data collection environment. So MVC numbers are normally only an estimate.
    Last edited by Edmund Cramp; June 18, 2021, 04:39 PM.

    Comment


    • #3
      Edmund Cramp,

      Thank you for your thoughts I will definitely look into the Gait analysis text.
      In your opinion what method would be better for normalizing muscle contraction while minimizing the effect of the data collection environment?

      Ricardo Sanchez

      Comment


      • #4
        Originally posted by Ricardo Sanchez View Post
        In your opinion what method would be better for normalizing muscle contraction while minimizing the effect of the data collection environment?
        We work to help users collect accurate data, not to tell them how to process it - I had a conversation with a respected clinical authority on gait analysis about 30 years ago about MVC processing and her point of view (which is now mine) was that while MVC is used and can generate reasonable results, the results depend on the way that the testing is performed. What is a maximum muscle contraction? Does the PT say, "Please push hard" or does she scream, "I'VE GOT A KNIFE" ... both generate a maximum contraction, but the second MVC recording might create signals twice the size of the first one. Everyone followed her instructions in the lab and generated consistent results that I have always respected. Her view was that while one environment can generate MVC consistent normalization, comparing the numbers to those generated in another researchers environment may show a undefinable difference.

        If you are collecting MVC data then you need to make sure that both the maximal and normal contraction data is processed identically - I would limit the bandwidth in both sets of data to eliminate both motion artifact and AC noise from the results when the MVC is calculated. While muscle contraction levels can be helpful, I see the timing of the physical motion generated by the contraction as being more significant.

        If you are looking at processing the EMG data then I would recommend storing the raw data first and then generating the processed results consistently and presenting both the raw data and the processed data together. In some circumstances EMG data can contain noise events (e.g. WiFi interference) that could affect the processed results - including the raw data helps everyone trust your results.
        EMG RFI issue.PNG
        Last edited by Edmund Cramp; June 20, 2021, 12:01 PM.

        Comment


        • #5
          As I understand it, the RMS more closely tracks the muscle tension (force or joint moment), which is generally what you are trying to detect. I don't know if there is some physical causal basis for this or whether it's just a coincidence. Perhaps someone could comment?

          Comment


          • #6
            Good to see you here again Chris, we added an RMS option to our EMG Analysis application, written for gait analysis years ago to meet a users request based on the same viewpoint.
            Ricardo, if you want to download the software from the link above, you can install it as a demo version and play with some of the demo C3D files to see the results of the different methods. The software gives you the option to play with different processing conditions and see Moving Average, Linear Envelope and RMS processing as well as other methods.

            Comment


            • #7
              [QUOTE=The software gives you the option to play with different processing conditions and see Moving Average, Linear Envelope and RMS processing as well as other methods.[/QUOTE]

              Edmund, this software looks like it's going to be so helpful. thank you so much for sharing, I ordered the text you recommended and I cant wait to get a better understanding of the analysis of EMG. I really appreciate the resources!

              Comment

              Working...
              X