Accuracy Improved Muscle Onset and Termination Detection in EMG Signal Processing Using STKEO

In electromyogram (EMG) analysis, accurate detection of onset of the muscle activity is crucial for various applications such as prostheses, robotics, disease diagnosis, and sports science. Therefore, an automated method that can accurately detect the beginning of muscle activity while reducing the noise is highly desirable. Teager-Kaiser Energy Operator (TKEO) is commonly used to achieve this objective which comes with a few drawbacks. Therefore, in this work, we propose the use of Shannon Teager Kaiser Energy Operator (STKEO) along with a new threshold value which is calculated considering the entire EMG signal rather than the signal baseline. The threshold values are formulated in such a way that they are independent of the preset value ‘h’, as the ‘h’ value varies among literature. The results suggest that STKEO with the new threshold value performs better than TKEO and other conventional processing methods, such as rectified EMG and linear envelope, in muscle-onset detection. The performance of the proposed method is evaluated in terms of error values by considering the statistical features of the processed EMG signals. From the results, it is observed that STKEO with the modified threshold resulted in a mean error percentage of 4.75 % while the error percentage of the TKEO method resulted in an error percentage of 14.77 %.