Volume 19, No. 2, 2022
Analysis Of Muscle Function For Postural Correction Using Surface EMG Signals
Suma K V , Lakshmi Shrinivasan , Varun C R , Nagendra B N
Abstract
When the muscle cells are activated electrically or neurologically, a kind of electric potential will be developed which is detected by a device named electromyography (EMG). The analysis of EMG signals can be utilized by Yoga therapists and physiotherapists who provide corrective therapy to the patients suffering from muscle related weakness and ailments. In this work, hardware module has been designed and developed to acquire EMG signals from healthy controls and subjects in need of posture correction. Four muscles namely right trapezius, left trapezius, right sternocleidomastoid muscle (SCM) and left SCM are identified for EMG signal acquisition. The features of EMG signals like root mean square value, signal power, signal frequency, motor unit potential and simple square integral are computed. Statistical measures like maximum, minimum, mean and standard deviation of these features are compared and are found to be significantly useful in analyzing the muscle activity. Support Vector Machine and Random Forest Decision Tree (RFDT) are used for classifying the signals. Classification is carried out on individual muscle signals and the accuracy is found to be highest for Right Trapezium and lowest for Right SCM. Further, classification based on combined features of all the four muscles is found to give accuracy of 88% and 81% for SVM and RFDT respectively. Based on accuracy results obtained using SVM and RFDT, the combination of features of all four muscles has been found to be satisfactory and comparative performance metrics are presented. This novel work can be helpful for treating the subjects with posture related muscular disorders.
Pages: 2796-2808
Keywords: EMG, Random Forest Decision Tree, Muscle Function, Support Vector Machine, Postural Correction