Elaine Corbett, BS
 |
|
Education
|
Graduate Student
Department of Biomedical Engineering
Northwestern University |
Research Interests
|
The goal of my research is to establish the relationship between prediction and performance accuracy for a variety of control algorithms for myoelectric arms driven by reinnervated muscle. Targeted muscle reinnervation enables the distinction of multiple imagined movements using pattern classification algorithms. In this field, algorithm success has traditionally been measured by prediction accuracy as calculated offline, where no performance feedback has been given. It is unknown whether prediction accuracy is an appropriate measure of algorithm performance during real-time, as user adaptation by means of feedback is not considered. A simple precision tracking task has been developed in order to assess performance of three algorithms: pattern recognition, nonparametric system identification and proportional control as used in conventional prostheses. Two surface EMG electrodes are placed on the forearm of healthy subjects, above wrist flexor and extensor muscles. This provides a good model of the signals available from a TMR subject with a transradial amputation. The direct measurement of wrist angle by means of a goniometer is also being considered. The user attempts to track a target in one dimension by controlling the velocity of a cursor displayed onscreen, where visual performance feedback is given. The prediction accuracy of the controllers is degraded in order to elucidate the relationship between accuracy and online performance. It is expected that for real-time online applications, algorithms may have robust performance accuracies over a range of sub-maximal prediction accuracies.
|
Contact Information
|
Suite 1328, Sensory Motor Performance Program
Rehabilitation Institute of Chicago
345 E Superior St
Chicago, IL 60611
USA
Ph: 312-238 1685
Email: e-corbett@northwestern.edu
|