Bayesian Filtering of Myoelectric Signals

• DOI: 10.1152/jn.00936.2006
• Publication Date: 2006-12-21T01:48:50Z
• Authors: Terence D. Sanger
• Keywords: SIGNAL (programming language)

Surface electromyography is used in research, to estimate the activity of muscle, prosthetic design, provide a control signal, and biofeedback, subjects with visual or auditory indication muscle contraction. Unfortunately, successful applications are limited by variability signal consequent poor quality estimates. I propose use nonlinear recursive filter based on Bayesian estimation. The desired filtered modeled as combined diffusion jump process measured electromyographic (EMG) random density exponential family rate given signal. estimated on-line calculating full conditional all past measurements from single electrode. gives that best describes observed EMG This yields results very low short-time but also capability rapid response change. approximates isometric joint torque lower error higher signal-to-noise ratio than current linear methods. Use significantly reduces noise compared algorithms, it may therefore permit more effective for control, neurophysiology research.