Spinal Interfacing via Muscle Recordings for Neuroprosthesis Control

Neuroprostheses may be used to substitute motor functions that are impaired as a result of injury or disease. Surface electromyography (EMG) is currently the most common clinical approach for extracting the user intent and for controlling prosthetics. In commercial control systems, the EMG amplitude is used for single degrees of freedom activation (direct control). However, this simple user interface neglects the full bandwidth of the neural drive to muscle underlying the EMG signal generation. Recent advances in technology for recording and processing EMG signals have been exploited to provide human-machine interfaces based on the neural output from the spinal cord, as decoded from the EMG. This chapter reviews technologies for detecting the activity of large populations of spinal motor neurons from muscle recordings with the aim of advancing clinically viable neuroprosthetics. We focus on invasive and noninvasive interfaces to record the muscle electrical activity and on deconvolution methods for extracting motor neuron discharge timings from these recordings. Finally, we report on two representative applications of this approach in the area of active prostheses (neurotechnologies for substitution) and pathological tremor suppression (neurotechnologies for neuromodulation).