Electromyography (EMG) records the electrical signals generated by muscles. To make sense of this information, it must be processed, which translates the complex electrical activity into a format that can be analyzed. This method decodes the electrical language of the muscular system, turning raw data into insights about how muscles function.
Understanding the Raw EMG Signal
The electrical activity measured in electromyography originates from Motor Unit Action Potentials (MUAPs), which are the electrical discharges from motor units. When these signals are captured, they appear as a complex oscillating waveform. This raw signal represents the sum of all MUAPs firing near the electrode, creating a signal with an amplitude in the range of 0 to 10 millivolts (mV).
There are two primary methods for acquiring this signal. Surface EMG (sEMG) uses non-invasive electrodes placed on the skin over a muscle, capturing a broad picture of superficial muscle activity. For more targeted data from deep or specific muscles, clinicians use intramuscular EMG, which involves inserting a fine-wire or needle electrode directly into the muscle tissue.
The raw EMG signal is not immediately usable because it is contaminated by noise and interference. Power-line interference introduces a consistent hum at 50 or 60 Hz from electrical equipment. Motion artifacts, caused by the movement of electrodes or cables, create low-frequency distortions. The signal can also be affected by biological crosstalk, where electrical signals from nearby muscles or the heart bleed into the recording.
Core Steps of EMG Signal Processing
The first step in processing a raw EMG signal is filtering to remove unwanted noise. A band-pass filter isolates the EMG signal’s energy, which for surface EMG is found between 20 Hz and 500 Hz. This process acts as a high-pass filter to remove motion artifacts and a low-pass filter to cut out high-frequency noise. A specific notch filter is also used to eliminate power-line interference at 50 or 60 Hz.
Following filtration, the signal undergoes full-wave rectification, which takes the absolute value of the signal to convert all negative amplitude values into positive ones. The raw EMG is an oscillating signal, and rectification ensures that all data points contribute to the analysis of the signal’s amplitude. This process makes the signal unipolar and prepares it for quantifying muscle activity.
The final step is smoothing, which transforms the rectified signal into a clean “envelope” showing the trend of muscle contraction intensity. This is achieved using a moving average filter, which calculates the average amplitude over a sliding window. Another method is calculating the Root Mean Square (RMS) value within successive windows, and both techniques produce a smoother waveform that is easier to analyze.
Extracting Meaningful Features
Once the EMG signal is cleaned and smoothed, specific metrics called features are calculated to quantify the muscle’s activity. These features are categorized based on whether they analyze the signal’s amplitude over time or its frequency components.
Time-domain features are calculated from the signal’s amplitude to measure the level of muscle activation or force. Common examples include the Mean Absolute Value (MAV) and the Root Mean Square (RMS). The MAV is an average of the rectified signal’s amplitude, while the RMS measures the signal’s power. These features are straightforward to calculate and are used in real-time applications for estimating muscle force.
Frequency-domain features are derived from a mathematical transformation of the signal to analyze its constituent frequencies. This analysis can reveal physiological changes in the muscle, such as fatigue, that are not apparent in the time domain. Features in this category include the Mean Frequency (MNF) and Median Frequency (MDF). These metrics are indicators of muscle fatigue, as the signal’s frequency content shifts toward lower values during sustained contractions.
Applications of Processed EMG Data
The data extracted from processed EMG signals has a wide range of uses. In clinical settings, EMG analysis is a diagnostic tool for identifying neuromuscular disorders. Neurologists examine EMG patterns for abnormalities that can indicate conditions such as muscular dystrophy, nerve damage, or amyotrophic lateral sclerosis (ALS). The signal’s characteristics can help pinpoint the nature and extent of the pathology, guiding treatment.
Advanced prosthetics use processed EMG signals to enable control of robotic limbs. Electrodes on a user’s residual limb detect muscle contractions, and the processed signals are translated into commands to control a prosthetic hand or arm. This allows for more natural control, enabling users to perform complex gestures.
In biomechanics and sports science, EMG helps analyze athlete muscle activation to optimize technique and prevent injury. Ergonomists use it to assess muscle strain in workplaces, improving safety and efficiency. The technology is also integrated into human-computer interfaces, allowing device control through gestures.