The recording properties of a multi-contact nerve electrode as predicted by a finite element model of the canine hypoglossal nerve

Most functional electrical stimulation (FES) systems rely only on unidirectional (i.e., efferent) activation of the target organ to yield therapeutic outcomes. For applications involving multi-fasciculated nerves, however, artificial sensors have exhibited limited results. As such, the flat-interface-nerve-electrode (FINE) is presented as a means of obtaining an effective closed-loop control system. To investigate the ability of this electrode to achieve selective recordings at physiological signal-to-noise ratio (SNR), a finite element model (JFEM) of a beagle hypoglossal nerve with an implanted FINE was constructed. Action potentials (AP) were generated at various SNR levels and the performance of the electrode was assessed with a selectivity index (0 ≤ SI ≤ 1; ability of the electrode to distinguish two active sources). Computer simulations yielded a selective range (0.05 ≤ SI ≤ 0.76) that was (1) related to the inter-fiber distance and (2) used to predict the minimum inter-fiber distance (0.23 mm ≤ d ≤ 1.42 mm) required for selective recording. The results of this study suggest that the FINE can record neural activity from a multi-fasciculated nerve and, more importantly, distinguish neural activity from pairs of fascicles at physiologic SNR.