The influence of digital filter type and normalisation method on surface EMG data during dynamic movement tasks

[Truncated] In the field of Electromyography (EMG), several different types of filters and normalisation methods have been employed to process the high frequency biological signal. The use of different types of filters and normalisation procedures for the signal processing of EMG data has added complexity when comparing these processed biological signals between studies. Additionally, there are numerous ways in which researchers analysed processed EMG data to obtain clinically relevant meaning. One example is co-contraction ratios, which like the EMG signal processing methods, vary extensively in the literature making comparisons between studies difficult at best.
To first understand the influence of normalisation method on the characteristics of a processed EMG signal, we processed EMG data using an underdamped Butterworth filter, then normalised these data using three commonly used methods in the literature; common clinically relevant dependent variables like mean total muscle activation (TMA) and directed co-contraction ratios (DCCR) were calculated from the processed data. The data was normalised using the peak muscle activation obtained from a combination isokinetic dynamometry trials, functional movement trials and dynamic calibration trials (COMB), peak muscle activation obtained from the functional movement trial only (FUNC), and average peak muscle activation obtained from straight line running trials (SLRm). From these analyses, results showed that the type of normalisation method employed significantly influences the clinical interpretation of a TMA and to a lesser extent DCCR estimates.