Segmentation of Surface Electromyography Signals: A Comparative Analysis of Time and Frequency Domain Methods

This study evaluates the efficiency of computational segmentation methods in electromyographic (EMG) signal analysis across two distinct exercise sets. Twenty participants were engaged, performing a series of isometric and isotonic exercises. The first set included four isometric handgrip exercises, while the second set consisted of four isometric exercises with measured weights and two isotonic exercises with weights. Out of the total, 15 registries from the first set and 18 from the second set were considered valid. The segmentation methods assessed were RMS, Integral, Variance, Mean, and Entropy. Entropy, with a beta factor of 5, demonstrated the highest segmentation efficiency of 0.88 for the first set and 0.75 for the second. The findings highlight the potential of the Entropy method in enhancing the accuracy of EMG signal segmentation, which is crucial for the development of biomechanical models and rehabilitation protocols.