Utilizing remaining voluntary muscle synergies to control FES elbow extension after spinal cord injury

Individuals with a C5/C6 spinal cord injury (SCI) have paralyzed elbow extensors, yet retain weak to strong voluntary control of elbow flexion and some shoulder movements. They lack elbow extension, which is critical during activities of daily living. This research focuses on development of a synergistic controller employing remaining voluntary elbow flexor and shoulder electromyography (EMG) to control elbow extension with functional electrical stimulation. We hypothesized that remaining voluntarily controlled upper extremity muscles could be used to train an artificial neural network (ANN) to control stimulation of the paralyzed triceps and provide functional benefits. Surface EMG was collected from SCI subjects while they produced isometric endpoint force vectors of varying magnitude and direction using triceps stimulation levels predicted by a biomechanical model. ANN'S were trained with the collected EMG and stimulation levels. The best set of muscles inputs was selected for each subject such that trained networks yielded low error, minimized the required number of EMG inputs, and effectively learned relationships between endpoint force and triceps stimulation. The synergistic controller increased the range of force vectors, provided discrete forces, and enabled an overhead reach task.