Non linear dynamical analysis of human skeletal muscles using surface electromyography signals

Abstract

Human body consists of more than 700 muscles. The dynamics of active muscles are highly non-linear. The non-linear dynamical behavior of muscles changes with movement and neuromuscular pathology. The occurrence of muscle fatigue is inevitable in the above conditions. Surface electromyography (sEMG) measures electrical signals from muscle non-invasively by placing electrodes on the skin. Many existing signal processing methods are available for the analysis of sEMG under fatigue conditions. newlineThis research aims to study muscle dynamics using sEMG signals and advanced non-linear techniques. For this, two datasets are considered in order to establish the need for normalization of maximum voluntary contractions (MVC). Hence, sEMG signals are acquired using two different experimental protocols one referenced to 50% MVC and the other under constant load until fatigue from Biceps brachii muscle of 45 and 50 healthy subjects respectively. The pre-processed signals are divided into ten equal segments. The first and last segments of the signal are considered as non-fatigue and fatigue, respectively. Further, the recorded signals are checked for their complex chaotic nature using different methods. The best fit for the NF and F segments is identified using sum of sine model. The NF and F segments are further transformed to Markov transition network (MTN) and Fuzzy recurrence network (FRN). Network measures are extracted from the segments to analyze their variations during fatigue. Finally, best features are identified for an automatic muscle fatigue detection system using four machine learning classification methods. newlineThe results indicated that the sEMG signals are complex and chaotic in nature. This may be due to the synchronized motor unit recruitment resulting in generation of self- similar patterns regularly until endurance. An attempt has been made to analyse the muscle elastance using network stiffness and is found to decrease during muscle fatigue.