Knee angle-specific EMG normalization: The use of polynomial based EMG-angle relationships

Abstract The normalization of EMG signals to those recorded during a maximal voluntary contraction provides a valid construct for comparisons of relative muscle activity. However, the length dependence of muscle activation and purported, substantial, muscle translocation and changes in muscle architecture during dynamic movements presents a need for joint angle-dependent normalization processes. The purposes of the present study were to: (1) quantify variations in muscle activity across a large ROM, (2) determine the accuracy with which fitted EMG-joint angle curves accurately characterized these variations, and (3) compare peak (EMG-P) and average (EMG-A) EMG amplitudes obtained during a countermovement leg extension when normalized to both absolute peak and joint angle-specific muscle activity. Fifteen subjects performed a large ROM (110°) isokinetic (30° s−1 ) leg extension from which EMG-joint angle relationships were derived using polynomial fitting of different complexities. Ten subjects also performed loaded countermovement leg extensions from which EMG signals were normalized using peak muscle activity and EMG-angle curves. EMG amplitude varied significantly over the ROM and the use of EMG-angle curves for signal normalization resulted in significantly greater EMG-P and EMG-A than those normalized using the absolute peak EMG. Higher-order polynomial fitting better matched the filtered EMG amplitudes. Thus, there is a strong rationale for using EMG-angle polynomial fits to normalize EMG signals for large ROM movements.