Recovery of Peripheral Nerve Signals Through Blind Separation

Abstract

Patients with neurological disorders can regain lost motor function through functional electrical stimulation (FES). In closed-loop prosthetic devices, neural signals can be obtained using cuff electrodes which have been shown to be stable for long-term recordings. Fibers in peripheral nerves are organized into fascicles, and usually contain both afferent and efferent signals. Therefore, peripheral nerves contain several independent signals that could be used as control signals. Several methods, such as microneurography and nerve cuff electrodes, have been used to selectively record from peripheral nerves. Selective recording can also be achieved by using nerve cuff electrodes placed around the nerve. The design of the FINE increases the proximity of a contact to a fascicle in a nerve by reshaping the latter and rearranging the fascicles within in it. Cuff electrodes have also been used to achieve fascicular selectivity based on, among other techniques on selectivity index (Yoo and Durand, 2005) In this presentation, I will demonstrate the selectivity of the FINE design on the hypoglossal nerve and investigate the feasibility of using independent component analysis (ICA) as a blind source separation (BSS) algorithm method to recover fascicular signals from simulated peripheral neural recordings.