Abstract
This paper presents a study of the use and accuracy of self-organizing maps (SOM) in classifying myoelectric signal properties. Myoelectric signals were obtained and classified for four upper-limb movements (elbow flexion, elbow extension, wrist pronation and wrist supination) and their force category. This was done for isolated actions as well as for multiple action sequences. The success of the developed SOM ranged from 92%-97% when determining the motion, from 81%-87% in determining the force category, and from 59%-96% in determining sequences of motions. These successes are encouraging for the continued development of this technique for use in controlling real-time complex motions in prosthetic devices.
Original language | English |
---|---|
Pages (from-to) | 147-152 |
Number of pages | 6 |
Journal | Biomedical Sciences Instrumentation |
Volume | 35 |
State | Published - 1999 |
Event | Proceedings of the 1999 36th Annual Rocky Mountain Bioengineering Symposium (RMBS) and 36th International ISA Biomedical Sciences Instrumentation Symposium - Copper Mountain, CO, USA Duration: Apr 16 1998 → Apr 18 1998 |
Keywords
- Myoelectric signal
- Pattern recognition
- Prosthetic limbs
- Self-organizing maps