Title :
[Copyright notice]
fDate :
Aug. 31 2010-Sept. 4 2010
Abstract :
The following topics are dealt with: Motor imagery-based brain computer interface (BCI) technology has motor rehabilitation as one of its main fields of application. The use of a BCI as a neuroprosthetic for paralyzed limb motor restoration implies normally absence of muscle activity. It is still an open question whether residual motor activity in healthy individuals or in patients causes a bias in the modulation of a motor imagery-based BCI control signal. Although the influence of electromyographical (EMG) activity in neck and cranial muscles upon BCI has been studied, not much has been said concerning the relevance of EMG activity arising from arm muscles. We therefore used a band motor imagery-based BCI system paradigm designed for motor rehabilitation coupling a BCI with an online driven robotic orthosis to compare different EMG activity detection methods regarding their influence in the resulting analysis of neurophysiological signals. Fourteen healthy subjects underwent four sessions in which they were asked to perform motor imagery task alone (receiving no feedback), motor imagery with (visual and proprioceptive) feedback, active movement, passive movement and rest. Six different EMG feature extraction methods were calculated and three different data time windows were used for muscle activity threshold definition. Three different electrode spatial distributions were utilized for removing the EMG artifacts: a) coming from all the electrodes on the arms, b) just the ones placed on the imagery side and c) just the ones on the healthy arm. We compared the different EMG rejection methods by calculating the number of trials deemed artifact-free by each method. In this paper we demonstrate that different EMG artifact removal methods lead to distinct partitions of the total available data, thus yielding different influence of the method used to remove EMG artifacts on task related artifacts regarding number of trials contaminated and the differences in trials rejec- ed using the different methods.
Keywords :
biomedical electrodes; brain-computer interfaces; electromyography; feature extraction; human-robot interaction; medical robotics; medical signal processing; muscle; neurophysiology; orthotics; prosthetics; EMG artifact removal methods; EMG feature extraction methods; EMG rejection methods; active movement; arm muscles; band motor imagery-based BCI system paradigm; cranial muscles; electrode spatial distributions; electromyographical activity; motor imagery-based BCI control signal; motor imagery-based brain computer interface technology; motor rehabilitation; muscle activity threshold definition; neck; neurophysiological signals; neuroprosthetics; paralyzed limb motor restoration; passive movement; proprioceptive feedback; residual motor activity; robotic orthosis; visual feedback;
Conference_Titel :
Engineering in Medicine and Biology Society (EMBC), 2010 Annual International Conference of the IEEE
Conference_Location :
Buenos Aires
Print_ISBN :
978-1-4244-4123-5
DOI :
10.1109/IEMBS.2010.5626543