While research shows that a patient’s success in using a myoelectric prosthetic arm is dependent on receiving effective training, current methods of training are not designed to effectively hold attention long enough for optimal training.
This study focused on evaluating a novel myoelectric training device, consisting of a toy car controlled by electromyograph (EMG) signals from the arm. http://en.wikipedia.org/wiki/Electromyography
The design of the circuit driving the myoelectric trainer was all based around an mBed embedded controller. Four EMG signals were pre-conditioned and then sampled, filtered, analyzed by the mBed to result in the motion of a remote controlled car.
The circuit was designed to accept four independent signals from a Noraxon MyoSystem 1200 EMG system through a twisted pair ribbon cable connector. The cable configuration was selected to minimize cross talk between the EMG channels. The signal for each input channel was then run through gain adjustment and offset adjustment circuits before being sent to the microcontroller.
These adjustments were under the experimenter’s control and were used during the calibration phase of the experiment to condition the EMG signals from the subject to fit the detection range of the microcontroller.
The microcontroller continuously runs its program to sample, filter, rectify, smooth, and normalize the incoming EMG signals, then analyze and compare the resulting signals to thresholds in order to determine what actions should be taken by the car based on the current amplitudes of the signals.
The resulting outputs were then sent to the car by closing relays connected to the joystick inputs on the original control board for the car. Subjects’ performance with the trainer was evaluated to determine the ability to provide experience with EMG controls.
Eight healthy adult subjects were taken through typical initial stages of myoelectric training, then asked to drive the car through a slalom course while the time, number of errors, and reversals required to complete the course were recorded, as well as the degree of difficulty subjects reported.
The learning induced by using the trainer was found to be statistically significant (p < 0.002), with subjects demonstrating dramatic improvements (> 49%) in performance.
To read this external content in full, download the report from the author archives at Virginia Commonwealth University.