Neural Signal Processing, Institute for Infocomm Research

Research Projects and Demonstrations

BCI-based Robotic Rehabilitation for Stroke

Collaborators

  • National Neuroscience Institute, Singapore
  • Tan Tock Seng Hospital, Singapore

Description

This project aims to develop the first neuro-rehabilitation system that combines non-invasive brain–computer interface (BCI) and robotic rehabilitation for patients suffering upper limb paralysis.


(click on image to see larger version)

With some 10,000 new cases of stroke each year in Singapore, up to 45% of patients do not completely recover bodily function. In addition, 85–90% of stroke survivors with severe upper-limb impairment do not regain full use of their arms or hands. Stroke rehabilitation has proven itself effective in restoring and improving function. Traditionally, it involves much human-to-human interaction between the therapist and patient. Recent research has shown encouraging preliminary results with new robotic-based rehabilitation techniques. Thus this project is designed to guide the user to perform rehabilitation exercises effectively, thereby maintaining strong motivation to progress. The use of BCI technology may potentially automate and facilitate the stroke-rehabilitation process and complement traditional therapies.

Clinical trials are currently being conducted at TTSH and NNI. Concurrently, the Neural Signal Processing Group is working together with the doctors in TTSH and NNI in the rehabilitation-engineering laboratory to improve the rehabilitation prospects of stroke survivors.

Upon successful neuro-rehabilitation, medical staff can quantify and serially monitor the intention, force, and motion the patient generates. The process will pave the way for a BCI-based robotic rehabilitation clinic where clients will receive consultation and training. By tapping the innovative system as part of their stroke-rehabilitation programme, patients will gain another therapy for rehabilitatio

Relevant Links

Motor Imagery-based Brain Computer Interface

Awards

  • 1st in International BCI Competition III Dataset 3a
  • 2nd in International BCI Competition III Dataset 4a

System Description

This is the technology that is used to control the movement of the robotic arm used in BCI stroke rehabilitation. The technology is based on recognition of the brain states when we carry out different types of motor imagery e.g. left arm moving, right arm moving. We research and develop semi-supervised methods to reduce the training effort required from the user, as well as feature extraction and classfication methods to improve the speed and accuracy of our technology.

Demo Description

Coming soon!

Relevant Links

BCI Competition III Results can be found here

Contact Us | ©2007 Neural Signal Processing Project, Institute for Infocomm Research