*9.1.1. Removing the slow-drift (or DC) component present in the signal*

Sometimes the signal involves a DC component that causes displacement of the baseline signal. This component is a common signal that has no relation with myoelectric activity. It can be the result of electrochemical phenomena between the electrodes and skin or the limitations of the amplifiers. An easy way to remove it is to calculate the average of all sampling points and shift the curve of the EMG result (high-pass filter) [12, 28].

### *9.1.2. Signal rectification*

Computational Intelligence in Electromyography Analysis – 402 A Perspective on Current Applications and Future Challenges

to the effects of sub-sampling (aliasing).

**8.3. Calibration** 

value, as shown in Figure 8.

**9. Mathematical processing** 

**9.1. Processing in the time domain** 

muscle force or human movement [25].

processing [1, 26, 28].

The sampling rate must also be very low compared to the frequency of signal variation due

An aliasing effect occurs whenever the sampling frequency is less than twice the highest

EMG recording is usually done at a maximum frequency of 500 Hz, and the sample should be at least 1000 Hz. To analyze muscle activity in the most comprehensive way possible, it is advisable to work with a sampling rate on the order of 2000 Hz, with the highest frequency

The measured physical magnitude is converted to voltage using a sensor or transducer, which is then applied to the A/D converter. Knowing the input range and resolution of the A/D converter, one can calculate the voltage of the converter input value from the digitized

Two types of processing are usually used in research: time domain processing, used when one is interested in the temporal analysis of EMG amplitude, and frequency domain

In order to process EMG signals in the time domain, there is a set of processing procedures for characterizing the curve and measuring the signal strength during muscle contraction. Having several kinesiological applications, EMG time domain analysis is often used in areas such as neuromuscular coordination, motor control, the relationship between EMG and

Sometimes the signal involves a DC component that causes displacement of the baseline signal. This component is a common signal that has no relation with myoelectric activity. It can be the result of electrochemical phenomena between the electrodes and skin or the limitations of the amplifiers. An easy way to remove it is to calculate the average of all

*9.1.1. Removing the slow-drift (or DC) component present in the signal* 

sampling points and shift the curve of the EMG result (high-pass filter) [12, 28].

frequency component of signal frequency, according to the Nyquist theorem [12].

component of the signal always limited by the low-pass filter [4, 12, 28].

**Figure 8.** Relationship of physical quantity to a digital signal [3].

Correcting the curve is an operation normally used to enable the subsequent integration of the signal, since it transforms a curve containing both positive and negative values (Figure 10) and a zero mean to a curve of only positive absolute values (Figure 11).

There are two ways to rectify the curve: eliminating the negative values (half-wave rectification), or reversing the negative values and adding them to the positive values (full wave rectification). Full-wave rectification has the advantage of maintaining all of the information contained in the signal, unlike half-wave rectification [5, 28].

### *9.1.3. Root-mean-square value of the signal*

The RMS is the amount of continuous signal able to contain the same amount of energy. It is mathematically defined as the square root of the mean of the squares of the instantaneous values of the signal [4, 12, 22, 23].

#### *9.1.4. Normalization of the signal in the time domain*

One problem when comparing different EMG signals has to do with differences in the duration of the various signals to be compared.

Normalizing means transforming, without changing the signal's structure, the duration differences into signals with the same number of samples. This can be done, for example, by taking the signal containing the lowest number of samples as a reference. An algorithm can be applied that determines, depending on the duration of each signal, the number of samples to be removed at certain intervals, reducing all signals to the same number of samples contained in the shorter of the two signals, and thus retaining the original forms [16].

#### *9.1.5. Amplitude normalization*

The EMG signal varies greatly upon comparison with recordings from the same individual or different individuals. The absolute value of the EMG signal thus provides little information, especially when dealing with signals from different individuals or the same individual at different times. One way to compensate for this limitation is to normaliz EMG amplitude curves. This technique consists of transforming the absolute amplitude values of the different curves to be compared into values relative to a reference EMG taken as 100% [4, 7, 15].

#### *9.1.6. Integral of the EMG signal*

The mathematical interpretation of the integral concept consists of determining the area enclosed by curve, whether an EMG or any other signal. In the case of the EMG, so that the result of integration is not zero, a rectified signal must be used. By integrating the EMG signal, a result that is proportional to the number of electrical impulses is obtained [3].

Computational Intelligence in Electromyography Analysis – 404 A Perspective on Current Applications and Future Challenges

#### *9.1.7. Filtering of the rectified signal*

The signals collected in real time in the original format are stored in files. After this phase certain mathematical processes are applied. The purpose of this processing is to make correction, i.e., to transform negative signals into positive signals. This is necessary to allow averaging of the analyzed signal, since if such correction is not performed, the average of the signals will be near zero. This is because the negative and positive are symmetrical. In the post-rectification, a 5 Hz low-pass filter can be run in order to have a signal wrap. The lower the value of this filter, the smoother the curve will be [27, 28].

Application of Surface Electromyography in the Dynamics of Human Movement 405

**9.2. Processing the frequency domain – Spectral analysis** 

been listed by various authors.

**10. Conclusion** 

**11. Future directions** 

mean velocity of muscle fiber conduction [3,4].

isometric contractions in volunteers with dysfunctions.

energy of noise from various undesirable sources [27].

should be increasingly directed to this specificity.

The EMG signal's frequencies are distributed between 1 and 500 Hz, with a great concentration between 20 and 250 Hz in the case of simple muscular activity. The distribution of energy at different frequencies (power spectral density) reflects the predominance of the low or high frequency components in the signal and has been used in kinesiological research. Factors that influence the spectral profile of the EMG signal have

EMG can be considered an overlapping of the action potentials of all the active motor units. The spectrum of EMG frequencies thus contains information about the characteristics of different fibers that contribute to the signal. Spectral analysis can provide information about the mean duration of the active fiber potentials, which in turn can be used to determine the

For dynamic sampling, active electrodes (with preamps) are less susceptible to artifacts or ambient noise, which can be observed when comparing them with signals collected during

EMG signals are affected by the anatomical and physiological properties of muscles, the peripheral nervous system and the instrumentation used to collect the signal. Thus it is

It can be said that signal processing begins, indirectly, as soon as the electrodes are placed. Electrode placement involves several factors that are decisive for the level and purity of the EMG signal to be collected, including: cleaning the skin, the amount and temperature of the conductive gel, the position of the electrodes and the signal-to-noise ratio, which expresses the balance between the energy of the signal generated during muscle contraction and the

Therefore, sEMG can be recommended as a tool for analyzing and interpreting electrical signals emanated during muscular contractions in both normal and pathological situations

Studies in the field of signal processing, especially, surface electromyography signals, have been widely used for understanding the dynamic motions by the fact that most human movements happening dynamically. Thus, processing in the field of time and frequency

Understanding the phenomena of depolarization of motor units, future research should be

related to the physiological, mechanophysiological and functional human movement.

and can be applied in the study of motor function and functional rehabilitation [4].

important to understand basic muscle functions to correctly record EMG signals [12].

**Figure 9.** A) original signal interference. B) rectified original signal [3].
