Analysis of algorithms for detection of pedaling intention in brain-machine interfaces

Abstract

The use of brain-machine interfaces in people who has suffered a cerebrovascular accident could help the rehabilitation process through the cognitive involvement of the patient. These interfaces translate the brain waves into commands to control the movement of an assistant mechanical device. However, the control of these devices should be more stable and achieve a higher accuracy. This work studies if algorithms, such as Stockwell or Hilbert-Huang transform, can improve the control of these devices, and if a personalization by subject or electrode configuration is desirable. Besides, through the analysis of five volunteers is determined that the motor intention can not be detected only by data acquired previously to the movement using desynchronized/synchronized related events. Therefore, it is needed to extend the time processing to the two seconds after the movement starting.