Identification Of Isometric And Dynamic Tasks Of The Upper Limb Based On High-Density Emg

Identification of tasks and estimation of volitional movement based on electromyography (EMG) constitute a known problem that involves different areas in the field of expert systems and particularly pattern recognition, with many possible applications in assistive and rehabilitation devices. The obtained information can be very useful to control exoskeletons or robots used in active rehabilitation processes. The emerging technology of high-density electromyography (HD-EMG) opens up new possibilities to extract neural information, and it has already been reported that the spatial distribution of HD-EMG intensity maps is a valuable feature in the identification of isometric tasks.This study explores the use of the spatial distribution of myoelectric activity and carries out a task identification during dynamic exercises at different velocities which are much closer to the ones commonly used during therapy. To this end, HD-EMG signals were recorded in a group of healthy subjects while performing a set of isometric and dynamic upper limb tasks. The results show that spatial distribution is a very useful feature in the identification not only of isometric contractions but also of dynamic contractions, so it can be very useful to improve the control of rehabilitation devices, making it more natural and permitting to adapt better to the user.