I am trying to learn about the muscle characteristics of a particular neurodegenerative disease:
1) What kind of information (i.e muscle weakness, Spasticity, fatigue) should I look out for in the sEMG that I have acquired?
2) Also, when speaking in terms of engineering, what kind of analysis (i.e frequency spectrum or power spectrum analysis or other techniques) do I have to conduct to obtain the following information:
i) Muscle Spasticity
ii) Muscle fatigue
iii) Muscle weakness
iv) Muscle activity
3) Does the EMG pattern vary from muscle group to muscle group within lower limbs?
4) Is there any literature, where I can find default EMG patterns of lower limb muscles, such as Gastrocnemius, Rectus femoris, Tibialis anterior, and Hamstring?
5) How do I make sure that I am not removing any useful information from EMG when I try to remove noise and artifacts?
6) Also how do I make sure I am not removing any irregular EMG pattern, which are due to pathological features of a disease and not a due to artifacts?
Hi Parthi,
1,2- Well, the answer is dependent on the disease, I mean the primary or secondary effects of that disease on a particular muscle. Any way, please find the files attached. Briefly, for fatigue analysis these features are suggested:
i) Linear methods
a) Time domain
- Root Mean Square (RMS)
- EMG envelope
- Zero Crossing Rate (ZCR)
- Turns per second
b) Frequency domain
- Mean frequency
- Median frequency
- Averaged Instantaneous Frequency (AIF)
- Dominant frequency
- Frequency at maximum power
- The power spectrum shape
- The total power
c) Time- Frequency (Scale) domain
- Wavelet analysis
d) JASA (Joint Analysis of EMG Spectrum and Amplitude)
ii) Nonlinear methods
- Higher Order Spectra Analysis (HOSA)
- Recurrence Quantification Analysis (RQA)
- Fractal dimension
- Correlation dimension
- Entropy, ...
3- In my opinion, patterns vary from muscle to muscle as well as person to person. It is natural, since different muscles perform different tasks and have various positions in relation to spine, head, trunk, etc.
4- Because of the variations mentioned in (3), I don't think so or at least I've not seen, yet. However, Kaplanis et al., have tried to compute the reference sEMG values for normal subjects recorded from Biceps Brachii muscle during isometric contraction in their article (P. A. Kaplanis, C. S. Pattichis, L. J. Hadjileontiadis, and V. C. Roberts, "Surface EMG analysis on normal subjects based on isometric voluntary contraction," Journal of Electromyography and Kinesiology, vol. 19, pp. 157-171, 2009). Moreover, EMG signals recorded during special kind of disease may have general patterns which differ from those of healthy ones, such as ECG or EEG signals for example during epilepsy (ictal) or cardiac diseases. But, these differences are just visual. However, as an engineer you may be interested in feature extraction from these visually different patterns to facilitate automatic machine learning.
Hope this information can be useful to you.
Bests,
Fariba
Dear Parthiban,
under the clinical point of view:
1- from sEMG you can receive informations about spasticity either in static or dynamic condition: during manual testing the slow and fast stretch of triceps surae allows you to detect the clonic answer of the muscles while keeping the knee bended at 90° of flexion or complete extension. You can perofrm three times a stretch manouvre and appreciate how long and strong is the clonic activity. In dynamic condition sEMG can give you data about flexion-relaxation phoenomenon in LBP patients (the literature is actually well describing very well the procedure). I avoid to talk about gait analysis, it will take too much time, but sEMG will enable you to observe hypoactivation or hyperactivity.
For the evaluation of muscle fatigue during isometric contraction, it's always useful to use mean frequency as a reliable parameter to study difference performance at different level and duration of disease.
cheers
1- You can extract limb posture information from sEMG, particular limb movement disease, and you may also retrieve other functional information depending on the particular area of interest. However, EEG signals can be more of use in ur case i thnk....
2- I thnk from Engineering perspective you may use multiple tools such as power spectrum, pattern recognition algorithms, wavelets, corelation, etc
3- Yes they do vary
4- YEs you may but the point is EMGs vary from person to person and also from target area to target area... so in a broader spectrum i thnk the correct answer is no. And thats why where you may find many databases of EEG and ECG on internet you will not be able to get hold of EMG database...
5- Well that is not any easy task may be a simulated EMG signal if generated and noises are added and after passing through your noise removal procces you calculate the SNR...
6- Remember there is always a compromise between noise removal and feature removal so point 5 is the answer to your question 6 but be very vigilent
Best of LUCK
@ Rabya: Thank you very much for u r suggestions. Sorry i wasn't clear about my 3rd question.
3. I know that the activation timing of muscles will vary, but what about their intensities??
4. You have any suggestions for getting hold of literature for EMG patterns of lower limb muscles???
Thanks
Parthi
Hi Parthi,
1,2- Well, the answer is dependent on the disease, I mean the primary or secondary effects of that disease on a particular muscle. Any way, please find the files attached. Briefly, for fatigue analysis these features are suggested:
i) Linear methods
a) Time domain
- Root Mean Square (RMS)
- EMG envelope
- Zero Crossing Rate (ZCR)
- Turns per second
b) Frequency domain
- Mean frequency
- Median frequency
- Averaged Instantaneous Frequency (AIF)
- Dominant frequency
- Frequency at maximum power
- The power spectrum shape
- The total power
c) Time- Frequency (Scale) domain
- Wavelet analysis
d) JASA (Joint Analysis of EMG Spectrum and Amplitude)
ii) Nonlinear methods
- Higher Order Spectra Analysis (HOSA)
- Recurrence Quantification Analysis (RQA)
- Fractal dimension
- Correlation dimension
- Entropy, ...
3- In my opinion, patterns vary from muscle to muscle as well as person to person. It is natural, since different muscles perform different tasks and have various positions in relation to spine, head, trunk, etc.
4- Because of the variations mentioned in (3), I don't think so or at least I've not seen, yet. However, Kaplanis et al., have tried to compute the reference sEMG values for normal subjects recorded from Biceps Brachii muscle during isometric contraction in their article (P. A. Kaplanis, C. S. Pattichis, L. J. Hadjileontiadis, and V. C. Roberts, "Surface EMG analysis on normal subjects based on isometric voluntary contraction," Journal of Electromyography and Kinesiology, vol. 19, pp. 157-171, 2009). Moreover, EMG signals recorded during special kind of disease may have general patterns which differ from those of healthy ones, such as ECG or EEG signals for example during epilepsy (ictal) or cardiac diseases. But, these differences are just visual. However, as an engineer you may be interested in feature extraction from these visually different patterns to facilitate automatic machine learning.
Hope this information can be useful to you.
Bests,
Fariba
@Partibhan
they vary in intensity, with variation in movement the frequency as well as signal energy varies...
The internet is ful of articles, however if you want me to provide somthng aproppriate let me see...
@Fariba
Ms. i dnt knw but whnever i try to downlod your pdfs an error prompts out
Hello Rabya,
I sent the articles to your email. I hope you will get them.
Best wishes,
- Badges
- Science topic
- Similar topics
- Biological Science
- Physiology
- Biosignals
- Biomedical Signal Processing