Temporal EEG analysis to detect the motor command generation
Recent work has shown that motor rehabilitation during the acute stages can decrease the effect of stroke. It is important to consider motor learning in the context of brain plasticity. The signal flow in the brain’s motor control system can be described as: a motor command generated in motor areas, that goes via subcortical and brainstem nuclei, to the spinal cord, and finally to specific muscles.
After muscle contraction, sensory feedback is transmitted to the somatosensory area in the cortex. This flow makes up one of the fundamental sensory-motor closed loops. However, stroke patients have difficulty learning specific motions because the feed-forward and feedback loops are damaged. If a correlation between motor command generation and the feedback signal to the somatosensory cortex can be achieved by artificial means, this opens up the possibility for re-forming the sensory-motor closed loop and thus, via learning, stimulate functional motor recovery.
Therefore, the recent development of the BCI (Brain Computer Interface) could be expected to be an important element of a system aiding reconstruction of the brain-body loop of stroke patients. We would like to develop electroencephalographic (EEG) analysis methods to detect the motor command generation.
The majority of EEG analysis has been based on either averaging raw signals across many trials, or examining a wide range of oscillatory processes. The event related (de)synchronisation (ERD/S) corresponding to suppression in the mu/beta band of the EEG is the most common one used for the detection of motor command generation. However, decomposition of the signal in the frequency domain is performed at the expense of time domain information, even though the time domain may contain useful and complementary information about the movement generation. Thus, we will develop time domain analyses to capture dynamical reorganisation of brain activity. The temporal dynamics of the relaxation of brain activity processes involved in motor activity can be captured by auto-correlation function, and we will analyse the relaxation time constant before the onset of motion, that is, how the brain activity becomes more correlated over time.
Project Research Group
Professor of Cybernetics, University of Reading
Lecturer of Robotics, University of Reading
Part IV Student, University of Reading