1. On the Suitability of Integrating Accelerometry Data with Electromyography Signals for Resolving the Effect of Changes in Limb Position during Dynamic Limb Movement.
- Author
-
Radmand, Ashkan, Scheme, Erik, and Englehart, Kevin
- Subjects
- *
ANALYSIS of variance , *ARM , *ELECTROMYOGRAPHY , *MULTIVARIATE analysis , *PROSTHETICS , *BODY movement , *ACCELEROMETRY - Abstract
Electromyogram (EMG) pattern recognition is an advanced signal-processing technique that has long been investigated as a method of improving the control of powered upper-limb prostheses. Several factors have recently been identified that affect the robustness of EMG pattern recognition, such as changes in limb position and performing dynamic activities. Some researchers have suggested that integrating accelerometry (ACC) data with EMG signals may provide the classifier with additional information about the limb position and therefore reduce the limb position effect. In these studies, however, it was assumed that the classifier was trained with data from all possible limb positions. This work investigates the suitability of combining ACC and EMG data when the training occurs only in a subset of tested positions. It is shown that, unless the classifier is trained in most of the possible positions, the integration of accelerometry data with EMG data can significantly degrade performance compared with EMG alone. In addition to the possible degradation of performance, because of the excessive training time, training the system in many positions is not clinically practical. Consequently, a dynamic training method that moves the residual limb through the regions of interest is introduced. This approach is shown to minimize training time while improving performance when combining ACC and EMG data. [ABSTRACT FROM AUTHOR]
- Published
- 2014
- Full Text
- View/download PDF