The Influence of Mechanical and Neural Coupling on Electromyography Signals of the Extrinsic Hand Muscles

Dexterous control of a myoelectric prosthetic hand has been the subject of decades of research and effort but remains an unsolved challenge. In part, it is because dexterous prosthetic control relies upon products from multidisciplinary fields: an articulated high degree-of-freedom (DOF) robotic limb, stable electrodes to capture high-resolution biosignals, the portable computational power to run control algorithms in real-time, and a decoder founded on a strong understanding of how the hand is controlled. Our scientific understanding of muscle activations coordinate to enable skilled and dexterous movements of the hand has not kept pace with the rapid development of technology which may be leveraged to develop a prosthetic hand. Dexterous control of the hand is a complex endeavor which requires simultaneous control over numerous muscles which span multiple joints. These muscles do not act independently and are coupled through mechanical means, such as shared tendons, and neural mechanisms, including diverging descending motor commands. Understanding how the coupling in the hand is present and manifested, particularly through the lens of electromyography, is a vital endeavor which is a necessary step towards development of dexterous prostheses. In this manuscript, I present my research that uses electromyography to characterize the mechanical and neural coupling within the extrinsic hand muscles. I begin with how the EMG activity of the extrinsic finger muscles, which span across the wrist, is affected by the posture of the wrist and contributes to wrist movements. In the next section, I demonstrate the neural coupling present during movements of the digits and develop a series of generalized models of coactivation and stabilization for the extrinsic finger flexors and extensors. Finally, using the EMG data collected I present the development of an optimization environment to test methods of EMG-to-activation and simulate the forward dynamics of a 17 DOF model of the hand to replicate dexterous movements. These results provide an insight on control of the hand which may prove useful for development of prosthetic control algorithms, and demonstrate a testing environment for EMG-to-activation signal processing protocols