PARAMETERIZING THE RELATIONSHIP BETWEEN EEG AND EMG DURING PRESCRIBED MOVEMENT

Abstract

Understanding the relationship between electrical activity within the cerebral cortex of the brain and muscular output is essential for advancing neuroprosthetic design and neuromotor rehabilitation. This thesis establishes and parameterizes the relationship between electroencephalography (EEG) and electromyography (EMG) signals during the prescribed contraction of the forearm and hand muscles in squeezing motions. First, it was confirmed that a reduction in alpha-band activity at the C3 EEG electrode corresponds to increased signal power in the flexor carpi radialis EMG electrode, upporting the hypothesis of cortical-motor coupling. Second, autoregressive (AR) models were constructed to characterize the spectral and stochastic properties of both EEG and EMG signals, achieving spectral flatness values of 0.91 and 0.96 inthe EEG and EMG models, respectively. Finally, a novel transfer function is proposed that defines the EEG-to-EMG system - yielding EMG-like output with a normalized spectral error of approximately 0.37 - and two methods of transfer function order selection are compared. These findings demonstrate the utility of system identification techniques in mapping cortical activity to muscular response, contributing to the development of interpretable and straightforward EEG-EMG models. Further, it may help to facilitate real-time control in brain-computer interfaces and related assistive technologies.