Algorithm for the detection of muscle activation in surface electromyograms during periodic activity

Area of Science:

• Biomechanics
• Motor Control
• Signal Processing

Background:

• Surface electromyography (EMG) is crucial for analyzing muscle activity.
• Quantifying muscle activation/deactivation during periodic motion is challenging, especially when maximum voluntary contractions are not feasible.
• Existing methods may lack precision in defining the initiation and cessation of muscle activity.

Purpose of the Study:

• To develop a simple surface electromyography (EMG) activation detection algorithm.
• To improve the numerical definition of muscle activity initiation and deactivation during periodic motion.
• To provide repeatable decomposition of EMG activity for statistical analysis.

Main Methods:

• Developed a simple algorithm using two interrelated, variable thresholds: percent amplitude and duration of a normalized cycle.
• Algorithm analyzes activation/deactivation periods as percentages of normal cycle parameters.
• Outputs include total percent activation per cycle, standard deviation of activity, and temporal indices of signal onset/offset.

Main Results:

• The algorithm demonstrated feasibility for repeatable decomposition of EMG activity.
• Initial coefficient of variation for percent activity per cycle was 0.24 (0.11).
• User selection between encompassing all non-baseline activity or peak activity only reduced the coefficient of variation to 0.16 (0.08).

Conclusions:

• A mathematically simple algorithm can effectively and repeatably decompose surface EMG activity.
• The algorithm's modifiable threshold parameter is essential for accommodating varying salient activity levels.
• This approach enhances the numerical definition of muscle activation/deactivation in periodic movements.