Muscle synergies for predicting non-isometric complex hand function for commanding FES neuroprosthetic hand systems

Myoelectric controlled neuroprostheses can restore hand function to mid-cervical level (C5/C6) paralyzed individuals through voluntary control. However restored functionality is limited due to the small number of available voluntary electromyographic (EMG) signals after paralysis. The purpose of this study was to determine whether dynamic hand function could be reduced to as few as three degrees of freedom using the time-invariant muscle synergy model thereby showing the feasibility of synergy-based neuroprosthetic control.

Task cross-validated, time-invariant synergies were derived from static hand postures and from dynamic functional task data collected from five able-bodied participants. The time-invariant synergies were extracted from EMG data in task cross-validation using non-negative matrix factorization.

Three functional synergies yielded significantly higher performance than chance (p   <  0.01) with 66.0%  ±  4.9% variance accounted for (VAF) compared to 42.5%  ±  4.4% VAF.

The results of this study, along with other studies showing continuous 3D EMG control, show the feasibility of a possible synergy-based controller for hand neuroprostheses.