AAHS Annual Meeting
Back to main AAHS site
Annual Meeting Home
Final Program
Past & Future Meetings

Back to 2015 Annual Meeting Program

Control of Upper Limb Prostheses By Activation of Motor Units in Targeted Muscle Reinnervated Patients
Aidan Roche, MBBS, PhD1; Hubertus Rehbaum, PhD2; Tamas Kapelner2; Ning Jiang, PhD2; Dario Farina, PhD2; Oskar C. Aszmann, MD, PhD1
1CD Laboratory for Restoration of Extremity Function, Division of Plastic and Reconstructive Surgery, Medical University of Vienna, Vienna, Austria; 2Department of Neurorehabilitation Engineering, Georg-August University, Göttingen, Germany

Introduction: The control of commercially available active prostheses relies on very simple systems for recording and processing the electromyographic (EMG) signals of remnant muscles. In these applications, the only information used for control is the strength of the muscle contraction. Targeted muscle reinnervation (TMR) improves this scheme by surgically isolating intuitive muscle contractions for prosthetic control, yet existing signal processing algorithms do not take full advantage of the information available. We hypothesized that more detailed information extraction from muscle signals, at the level of individual motor units, would provide increased numbers of useable control signals for reliable use of prostheses without further surgical intervention.

Methods: Using a novel pattern recognition system for extracting control information from the global EMG signal, we have developed a technique that identifies individual motor unit behavior. This method relies on multi-channel EMG recording and decomposition of muscular electrical activity into the muscle fiber action potentials and the innervating nerve pulses. The nerve pulses can then be used to extract the patients’ intent and thus identify the motor tasks they wish to execute. We applied this method to signals recorded from 3 glenohumeral patients who underwent TMR. Outcome measurements were recorded as percentage classification accuracy where the system’s predictions were compared to patients’ intent.

Results: While completing 8 (subject 1), 10 (subject 2), and 12 (subject 3) motor tasks of the arm, wrist and hand, the patients’ intent could be correctly identified with an accuracy of >98% using this novel approach. The same tasks could only be classified with an accuracy of approximately 85% when using EMG activity as a global signal.

Conclusions: These results demonstrate the high accuracy of this novel approach based on motor unit behavior and are promising for improving pattern-recognition control of active prostheses in TMR patients.

Back to 2015 Annual Meeting Program
© 2020 American Association for Hand Surgery. Privacy Policy.