Delsys EMG Workshop Analysis

The goal of this Flash Lab was to examine the EMG signal of the muscle fiber between the thumb and the pointer finger of the right hand of the test subject (me) using the Delsys Tringo sensor and extract the meaning of the EMG data acquired.

In the image above, it illustrates the results to find the sampling period and sampling frequency of the signal analytically. The sampling period was obtained via simply dividing the total time duration (ms) by the # of samples (13.176/25382 = 0.000519108). The sampling frequency was obtained and confirmed via dividing 1 by the sampling period (1/0.000519108 = 1926.381299). The reported sampling frequency by the Trigno sensor data is rounded and may not be exact which is explains the difference.

Above is an image of the calculated average and RMS data of the chosen segment. The average of the segment illustrates the typical point in time and voltage at which the signal lies. The root mean square (RMS) is basically the square of the signal where first the average is obtained then, the square root. It represents the square root of the power of the EMG signal for a given period of time. Furthermore, these are both time domain variables because the amplitude of the signal is measured as a function of time. The RMS will provide the most insight of the amplitude of the signal better than the average.

Above is an image of the FFT (Fast Fourier Transform) of the segment along with the translation into amplitudes. To obtain the frequency, each point number is divided by the total amount of time of the segment (ex. point 1 is 1/1.062614136 = 0.941075378 Hz). The frequency values are recorded in Hertz (Hz) while the amplitude is a value obtain via using the IMABS() function in excel which returns the absolute value of a complex number in the for x +yi or x + yj. This function translates the FFT complex numbers into amplitude values which allow for a plot of the frequency versus magnitude spectrum of the EMG signal segment to be produced.

Above is images of the Frequency Magnitude Spectrum of the EMG signal, on the y axis is the amplitude of the signal while on the x axis is the frequency of the signal. The left figure shows the original spectra, which is graphed with its mirror image that causes it to appear "folded". The upper half of the magnitude and phase spectra are mirror images of the lower half and therefore redundant. So, the figure to the right illustrates the first half of the magnitude spectra which is all that is needed for frequency and magnitude analysis. The spectrum above fs/2 (in this case, fs is 1926 Hz so half the specta would be 963 Hz) is a mirror image of the lower half of the spectrum. When we sample a continuous signal, we effectively multiply the original signal by a periodic impulse function (with period fs) and that multiplication process produces all those additional frequencies. Even though the negative frequencies are mathematical constructs, their effects are felt because they are responsible for the symmetrical frequencies above fs/2.

As seen in the image, a few key frequencies and their amplitudes are labeled on the graph. The usable energy of the EMG signal is dominant between 50 - 150 Hz which is primarily where this signal lies and should be the signal bandwidth. Noise can be observed in the signal around 61.17 Hz along with 48.94 Hz which are likely ambient noise. Any frequencies between the 0 to 20 Hz range are likely motion artifacts. Moreover, noise is also observed after 143.98 Hz since that is out of the bandwidth of the signal; however, some frequencies of the EMG signal can lie between 0 - 500 Hz. The dominant frequencies are 95.99 Hz, 118.58 Hz, and 143.98 Hz since these frequencies present the highest magnitude of amplitude value in the signal spectra. It is difficult to identify the noise artifacts in the time domain since the data comes in raw and unfiltered; however, in the frequency domain one can quickly determine the dominant frequencies and pin point the outliers in the data which appear to be out of the desired bandwidth for the EMG signal.

Noise can be described as any portion or aspect of the output signal which is undesirable and may possibly mask the true signal of interest. Noise can have many sources and interpretations; it can come from external radiated sources (such as “line interference”), it can be caused by electrical disturbances intrinsic to the recording environment (such as motion or stimulus artifact), and it can be caused by the nature of the recording devices themselves. Types of noise include; inherent noise in electronics components in the detection and recording equipment, ambient noise, motion artifacts, and inherent instability of the signal. The design of the EMG equipment, the establishment of a noise-free recording environment and the methodologies for using the EMG equipment must be carefully considered so that the EMG signal may be recorded with high fidelity and so that the signal-to-noise ratio is maximized.

The electrode type and amplifier characteristics play a crucial role in obtaining a noise-free signal. For kinesiological EMG there are two main types of electrodes: surface and fine wire. The surface electrodes are also divided into two groups. The first is active electrodes, which have built-in amplifiers at the electrode site to improve the impedance (no gel is required for these and they decrease movement artifacts and increase the signal to noise ratio). The other is a passive electrode, which detect the EMG signal without a built-in amplifier, making it important to reduce all possible skin resistance as much as possible (requires conducting gels and extensive skin preparation). With passive electrodes, signal to noise ratio decreases and many movement artifacts are amplified along with the actual signal once amplification occurs. The advantages of surface electrodes are that there is minimal pain with application, they are more reproducible, they are easy to apply, and they are very good for movement applications. The disadvantages of surface electrodes are that they have a large pick-up area and therefore, have more potential for cross talk from adjacent muscles. Additionally, these electrodes can only be used for surface muscles. Fine wire electrodes require a needle for insertion into the belly of the muscle. The advantages of fine wire electrodes are an increased band width, a more specific pick-up area, ability to test deep muscles, isolation of specific muscle parts of large muscles, and ability to test small muscles which would be impossible to detect with a surface electrode due to cross-talk. The disadvantages are that the needle insertion causes discomfort, the uncomfortableness can increase the tightness or spasticity in the muscles, cramping sometimes occurs, the electrodes are less repeatable as it is very difficult to place the needle/fine wires in the same area of the muscle each time. Additionally, one should stimulate the fine wires to be able to determine their location, which increased the uncomfortableness of using this type of electrode. However, for certain muscles, fine wires are the only possibility for obtaining their information.

There are many other amplifier characteristics which should be noted. The first of which is the signal to noise ratio. This is the ratio of the wanted signal to the unwanted signal and is a measure of the quality of the amplified signal. The higher the ratio, the greater the noise reduction. Electrodes with on-site pre-amps (miniaturized and at the site of the electrode) are some of the best at providing a very large signal to noise ratio. The gain of the amplifier is also important. This is the amount of amplification applied to the signal and it should be sufficient enough to have output amplitude at 1.0 volt. Another important characteristic of the amplifier is the bandwidth. This is simply the range of the collectable frequencies of the amplifier, and one wants this high enough to reject the low frequency movement artifacts and low enough to attenuate the signal as little as feasible. This means in general, one should be collecting in a range from 0 Hz to 600 Hz for surface electrodes and 0 Hz to 1,000 Hz for fine wire electrodes. Using the Nyquest Theorem, this means that one must sample at a minimum of 1,200 Hz for surface electrodes and 2,000 Hz for fine wire electrodes in order to assure capturing the entire signal. Once the signals have been recorded, then one could use a 10-15 Hz high-pass filter to eliminate the movement artifacts (some prefer to use an analog filter on the front end, but I prefer to filter movement artifacts after collection).

The goal is to educate others on what the importance of EMG analysis and what the EMG signals tell you about how the muscles are performing. What can you learn about performance from EMG?

Reflection: How would you summarize what you have learned about EMG signal analysis? In your reflection include observations about the following:

a) Data and file formats: how did you have to handle them?

b) Working with signals in the time and frequency domain, advantages/disadvantages of each.

c) Challenges in working with real biomedical data.

d) How could you extend this activity and relate it to something with direct impact to yourself?

Electromyography (EMG) is the study of muscle function through analysis of the electrical signals emanated during muscular contractions. Electromyography is often abused and misused by many clinicians and researchers. Many times even experienced electromyographers fail to provide enough information and detail on the protocols, recording equipment and procedures used to allow other researchers to consistently replicate their studies. Electromyography is measuring the electrical signal associated with the activation of the muscle. This may be voluntary or involuntary muscle contraction. The EMG activity of voluntary muscle contractions is related to tension. The functional unit of the muscle contraction is a motor unit, which is comprised of a single alpha motor neuron and all the fibers it enervates. This muscle fiber contracts when the action potentials (impulse) of the motor nerve which supplies it reaches a depolarization threshold. The depolarization generates an electromagnetic field and the potential is measured as a voltage. The depolarization, which spreads along the membrane of the muscle, is a muscle action potential. The motor unit action potential is the spatio and temporal summation of the individual muscle action potentials for all the fibers of a single motor unit. Therefore, the EMG signal is the algebraic summation of the motor unit action potentials within the pick-up area of the electrode being used. The pick-up area of an electrode will almost always include more than one motor unit because muscle fibers of different motor units are intermingled throughout the entire muscle. Any portion of the muscle may contain fibers belonging to as many as 20-50 motor units.

EMG signal analysis is important to help diagnose or rule out a number of conditions such as: muscle disorders like muscular dystrophy or polymyositis. Any diseases that affect the connection between the nerve and the muscle such as, myasthenia. It can also be used by researchers involved in biomechanics, motor control, neuromuscular physiology, movement disorders, postural control, and physical therapy. Therefore, it is crucial to analyze EMG data to help determine how well the muscles are performing along with evaluating and recording the electrophysiological activity produced by the persons muscles. An abnormal EMG result can identify a problem in an area of muscle activity which helps with diagnosing medical conditions. EMG can reveal nerve dysfunction, muscle dysfunction or problems with nerve to nerve muscle signal transmission. Therefore, this test can help determine if the muscles are responding the right way to nerve signals or if they may be damaged or diseased which can impede their performance.

EMG signal analysis is very complicated. Often noise may emanate from various sources which can affect the results of the frequency magnitude spectra produced. The electrode type and amplifier characteristics play a crucial role in obtaining a noise-free signal as described in the section before this one. Therefore, it can be difficult to work with real biomedical data as a multitude of factors can impact the result. Everything from how the electrodes designed and how the sensor is designed to the actual analysis of the signal can be extremely challenging. The original data is obtained via recording the time in milliseconds and milliVolts; however, any noise can impede from quick and easily interpretation of the EMG signal. As seen in the time domain analysis it is easy to see the signal recorded in real time; however, it can be difficult to quickly identify noise and key frequencies. Therefore, it is best to break down the EMG data into window segments and analyze segments of the data. The time domain analysis has the advantage of quick interpretation of the strength of the EMG signal and if the sensors, electrodes and amplifier electronics within the sensor are correctly places and picking up strong voltage samples of the data. However, it is difficult to analyze because it does not detail any information about the power density of the signal. Therefore, the frequency domain analysis is used to interpret the time domain data and provide a better analysis of how strong the EMG signal really is coming. The frequency domain allows for quick and clear determinations of the strength of key frequencies based on how high their amplitude is; moreover, this method displays how much of the signal exists within a given frequency band concerning a range of frequencies. The time domain analysis only displays the changes in a signal over a span of time. This activity could be expanded and related to myself to determine the performance of my muscles. Furthermore, by placing the sensor over certain muscles on my body I can determine if the muscles in my body are all functioning properly or if I have caused any injury. This can be crucial when working out as muscles can be strained and damaged with over use along with excessive weight being lifted. This will allow me to determine how to prevent injury along with improve my muscular endurance by understanding how the signal changes with different exercises.