Teager–Kaiser energy operator signal conditioning improves EMG onset detection

Abstract

Accurate identification of the onset of muscle activity is an important element in the biomechanical analysis of human movement. The purpose of this study was to determine if inclusion of the Teager–Kaiser energy operator (TKEO) in signal conditioning would increase the accuracy of popular electromyography (EMG) onset detection methods. Three methods, visual determination, threshold-based method, and approximated generalized likelihood ratio were used to estimate the onset of EMG burst with and without TKEO conditioning. Reference signals, with known onset times, were constructed from EMG signals collected during isometric contraction of the vastus lateralis (n = 17). Additionally, vastus lateralis EMG signals (n = 255) recorded during gait were used to evaluate a clinical application of the TKEO conditioning. Inclusion of TKEO in signal conditioning significantly reduced mean detection error of all three methods compared with signal conditioning without TKEO, using artificially generated reference data (13 vs. 98 ms, p < 0.001) and also compared with experimental data collected during gait (55 vs. 124 ms, p < 0.001). In conclusion, addition of TKEO as a step in conditioning surface EMG signals increases the detection accuracy of EMG burst boundaries.