A patient-driven rehabilitation system to improve upper limb amputee outcomes. The goal of our research program is to develop technological solutions that restore function for upper extremity amputees. For these individuals, one of the most promising developments in the past decade has been the introduction of fully dexterous terminal devices. However, an effective and intuitive control strategy has remained elusive. Myoelectric control works well for one or two degree-of-freedom devices, but is not effective when dealing with a large number of movement classes (such as when attempting to discriminate amongst """"""""hand open"""""""", """"""""index finger point"""""""", """"""""fine pinch"""""""", and so on). With this background, several groups pioneered the use of pattern recognition algorithms. Near perfect results have been demonstrated in the laboratory setting, but to date such performance has not been translated into real-world clinical impact. We believe that a critical missing piece is effective patient training. During Phase I, we developed such a methodology. MyoTrain v1.0 is a desktop computer based virtual-hand training system, including hardware and software components. The hardware consists of a compressive silicone cuff embedded with eight electrode pairs, an EMG amplifier, and a data-acquisition card contained in a desktop PC. The software includes a virtual prosthesis model and multi-channel EMG pattern-recognition software. The system was evaluated in a study involving four trans-radial amputees. Control of nine different movement classes was achieved with a Movement Completion Percentage of 99.0% (versus 70.8% at baseline). Furthermore, excellent accuracy was maintained over 8 hours without requiring any classifier retraining. We now propose a Phase II effort involving translation of our clinical prototype into a medical device cleared by the United States Food and Drug Administration: MyoTrain v2.0. As part of this journey from lab to clinic to home, our Phase II plan involves the development of modules for telerehabilitation as well as for immersive virtual reality and task-based training. This work is based on direct feedback we have received from amputees and clinicians using MyoTrain v1.0, and we will continue to interact with these groups during development of MyoTrain 2.0 to yield the best possible clinical outcome. We will then conduct a study over a period of two months to evaluate clinical impact. Our primary outcome measure is the Southhampton Hand Assessment Procedure, but we will also employ the Box and Blocks test as well as other measures including three different validated survey instruments. We believe that our Phase II program will yield a device with FDA 510(k) Premarket Clearance which will, for the first time, enable upper-extremity amputees to unlock the full potential of today's dexterous prosthesis.
The loss of an arm is devastating to almost any individual. In this project, we will develop and launch a rehabilitation system that will enable individuals to more easily transition to the use of a dexterous prosthesis with intuitive control.
| Betthauser, Joseph L; Hunt, Christopher L; Osborn, Luke E et al. (2018) Limb Position Tolerant Pattern Recognition for Myoelectric Prosthesis Control with Adaptive Sparse Representations From Extreme Learning. IEEE Trans Biomed Eng 65:770-778 |
