Muscle control using proportional detection of electromyographic signals

After a spinal cord injury an individual may retain some natural, neurological control over muscles although the signal may not be large enough to promote actual muscle contraction. Herein electromyographic signals, collected at the skin surface, provide input to a proportional detector comprising nonlinear filters and a multi-level adaptive thresholder. This investigation develops a methodology for determining a muscle contraction that indicates levels of recruitment. Further, the results are both robust to random nerve firings and adaptive to fatigue; and are validated in a controlled laboratory setting. In this manner, the central nervous system's own integration system may be used to control and coordinate muscles being artificially stimulated.<>