Performance comparison of ICA algorithms for isometric hand gesture identification using surface EMG

There is an urgent need for developing a robust technique that can identify small and subtle hand and other body movements with applications in health, rehabilitation and defence. Surface electromyogram (sEMG) is a measure of the electrical activity of the muscles and a measure of the strength of muscle contraction. While this may be a good measure of the actions and gestures, this is unable to identify small variations in the muscle activity, especially when there are number of simultaneously active muscles. Independent component analysis (ICA) is a statistical based source separation technique that has been shown to be suitable for the decomposition of signals such as sEMG and been shown to improve the ability of sEMG to identify small variations in muscle activity. ICA algorithms using multivariate statistical data analysis technique have been successfully used for signal extraction and source separation in the field of biomedical and statistical signal processing. Recent research has resulted in the development of number of different ICA technique. While there are some researchers who have compared their techniques with the existing methods for audio examples, there is no comparison of performance between ICA algorithms for biosignal applications such as surface electromyography (sEMG) applications. With ICA being the feasible method for source separation and decomposition of biosignals, it is important to compare the different techniques and determine the most suitable method for the applications. This paper has studied the performance of four ICA algorithms (FastICA, JADE, Infomax and TDSEP) for decomposition of sEMG to identify subtle hand gestures. Comparing several ICA algorithms, it is observed that an algorithm based on temporal decorrelation method (TDSEP) which is based on the second order statistics gives the best performance.