Myoelectrical control / BCI
Neuromuscular disorders such as amyotrophic lateral sclerosis or spinal cord injury disrupt the pathways through which the brain normally transmits motor commands to the peripheral organs. These diseases may cause personal and social burdens for the patients. Although modern rehabilitation prosthetics partly resolve some of these issues, discovering effective and user-friendly assistive devices is of great interest. Advances in hardware and signal processing techniques have narrowed the gap between science fiction and reality by effectively extracting movement related signals from neural recordings, bypassing the broken pathways. A brain computer interface (BCI) translates signatures of neural activity related to motor intention or movement imagination to decision signals to drive external motor prosthetics or rehabilitation devices. Adaptive and intelligent BCI technology merges the outer world and users’ own cognitive space and extends the boundaries of their body.
Transforming the conventional pre-programmed BCI systems into intelligent interfaces which are autonomously optimized with respect to user performance is of fundamental importance. In an ideal ``co-adaptive'' scenario, where the subject and the interface are collaboratively optimized for an end goal, a measure of motor control that reflects neural or behavioural accuracy of the user should be integrated into the interface. Recent advances in computational neuroscience, in particular the development of the Optimal Feedback Control (OFC) model, have provided new insights into potential methods for achieving co-adaptive control in BMIs. However, at present experimental electrophysiological evidence for OFC is lacking. My current research is therefore addressing this gap in our knowledge.
The project will use both an invasive BCI and a related myoelectric-controlled interface (MCI) that can be implemented with human subjects using surface electromyogram (EMG). Since the firing rates of many cortical neurons are consistently correlated with EMG over a wide range of motor tasks MCIs can approximate some features of a BCI without the need for invasive recordings. This will allow the experimental methods to be optimised and validated in humans before invasive procedures are performed.
People involved in this project
Sponsors of this study
Back to the Projects