DocumentCode :
122471
Title :
Brain-computer interface for neurorehabilitation: Looking beyond upper limbs
Author :
Cuntai Guan ; Huijuan Yang ; Kai Keng Ang ; Kok Soon Phua ; Juanhong Yu ; Chuanchu Wang ; Chua, Kee-Chaing ; Chew, Effie
Author_Institution :
Inst. for Infocomm Res., Singapore, Singapore
fYear :
2014
fDate :
17-19 Feb. 2014
Firstpage :
1
Lastpage :
1
Abstract :
With deeper understanding and appreciation of the roles of Brain-computer interface (BCI) in assisting stroke survivors to restore motor function by inducing activity-dependent brain plasticity through Hebbian learning, more and more studies in applying BCI for stroke rehabilitation have been conducted. Previous studies mainly focused on upper limb rehabilitation, typically by combining BCI with a mechanical feedback device (robotic arm or haptic knob) or functional electrical stimulation (FES). In our lab, in collaboration with clinicians in Tan Tock Seng Hospital, National Neuroscience Institute and National University Hospital, we have conducted three clinical studies involving more than 60 hemiplegic stroke patients to perform upper limb rehabilitation. In these studies, we observed statistically and clinically significant improvement in patients´ upper limb recovery comparing their post-with pre-rehabilitation assessments. Neural imaging also shows statistically significant enhancement in functional connectivity. Learning from the upper limb rehabilitation, we are interested in applying BCI for the rehabilitation of lower limb, which is equally important for the improvement of a patient´s quality of life, but more challenging compared with that for upper limb due to less alternatives available. In this talk, we present a study on the detection of motor imagery of brisk walking, aiming at developing a training system for lower limb rehabilitation. We are particularly interested in identifying the most relevant channels and frequency bands with regard to the detection of motor imagery of brisk walking from the EEG data when a subject imagines brisk walking. Specifically, we propose to select the most informative channels and frequencies by jointly maximizing the mutual information between the laplacian derivatives of power features and class labels, and minimizing the redundancy between the to-be-selected features with those already selected. Evaluated on heal- hy subjects, the results demonstrated that the most frequently selected channels were mainly located at the premotor cortex, supplementary motor area, dorsolateral prefrontal association cortex and posterior somatosensory association cortex. A clinical study using this system for lower limb rehabilitation is on-going in National University Hospital, Singapore.
Keywords :
biomedical equipment; brain-computer interfaces; electroencephalography; gait analysis; medical disorders; medical image processing; neurophysiology; patient rehabilitation; plasticity; statistical analysis; BCI; EEG data; Hebbian learning; Laplacian derivatives; activity-dependent brain plasticity; brain-computer interface; brisk walking; dorsolateral prefrontal association cortex; frequency bands; functional electrical stimulation; hemiplegic stroke patients; informative channels; lower limb rehabilitation; mechanical feedback device; motor function; motor imagery detection; neural imaging; neurorehabilitation; premotor cortex; prerehabilitation assessments; somatosensory association cortex; statistical significant enhancement; stroke rehabilitation; stroke survivors; supplementary motor area; upper limb rehabilitation; Brain-computer interfaces; Educational institutions; Electroencephalography; Hospitals; Legged locomotion; Neuroscience;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Brain-Computer Interface (BCI), 2014 International Winter Workshop on
Conference_Location :
Jeongsun-kun
Type :
conf
DOI :
10.1109/iww-BCI.2014.6782556
Filename :
6782556
Link To Document :
بازگشت