A method to test reliability and accuracy of the decomposition of multi-channel long-term intramuscular EMG signal recordings

Abstract More and more research groups use intramuscular electromyogram (EMG) recordings and there is a need for evaluating decomposed signals with respect to reliability and accuracy. We propose and tested a method to evaluate reliability and accuracy of the decomposition procedure of real, multi-channel long-term EMG signals. The method is based on a comparison of the decomposition results of the original signals with the firing distribution of the rotated signals. The rotation is achieved by reversing the original signal so that the last sample becomes the first and the first sample the last. A number of performance parameters were defined to compare the signals. The method was applied on 76 intramuscular long-term EMG signals that were automatically decomposed by our decomposition software (EMG-LODEC). The EMG signals contained between 0 and 37923 motor unit action potentials. The method enabled us to test the quality of the decomposition results of intramuscular signals and to scrutinize the non-reliable part of the decomposition to ensure that the results were accurate. Achieved reliability was good in approximately 50% of the recordings. Only four were clearly insufficient. Relevance to industry Work-related shoulder-neck pain is a major health risk in computer operators. To understand the physiological mechanisms behind the development of these disorders, EMG recordings of some minutes up to several hours must be accurately decomposed. This paper presents a method to test reliability and accuracy of the decomposition procedure of measured intramuscular signals of long duration. The method seems to be suitable for evaluating decomposition results of measured EMG signals and constitutes a useful tool for EMG researchers to test their intramuscular recordings.