Impedance and Signal Quality Measurements for Microelectrode Arrays Implanted Chronically In Feline Spinal Nerves

The spinal nerves, consisting of ventral roots and dorsal root ganglia, are a novel target for motor and sensory interfaces for neural prosthetic limbs. By implanting floating microelectrode arrays (FMA) into the spinal nerves, it is possible to record single unit activity from motor axons in the ventral roots (VR) or microstimulate sensory fibers in the dorsal roots and dorsal root ganglia (DRG). The ventral roots remain a lesser-explored electrophysiological target for motor control of a neuroprosthetic device. Long term acquisition of high quality neural recordings is necessary for neuroprosthetic applications and electrical interface impedance measurements are often used as a diagnostic tool to evaluate electrode integrity. However, it is not well understood how electrode impedance affects the quality of single unit recordings, particularly in the spinal nerves where few chronic recording studies have been performed.\ud This study characterizes ventral root recordings by examining signal-to-noise ratios (SNR) and single unit yield for chronically implanted microelectrode arrays in cats. Two 32-channel floating microelectrode arrays (FMA, MicroProbe, Inc.) were implanted into the L6 and L7 ventral roots of 9 cats during sterile surgery. Single unit recordings were performed during treadmill locomotion and under anesthesia. All recorded units were classified as motor or sensory by heuristic metrics after all data was manually spike sorted. For all motor-related units, SNR and single unit yield were calculated for each electrode. The SNR and yield measures were grouped by electrode site size to examine the effect of site size and electrode impedance on recording quality. The electrode site exposures ranged from 25 to 160 µm, yielding initial impedance values in the range of 50-500 kΩ. As expected, electrode impedances were inversely correlated with site size. However, SNR and yield did not differ significantly across this wide range of electrode sizes; peaks in both probability of unit detection and median SNR values did not consistently fall within a particular range of impedances. Both SNR and yield decay over time, as expected, but all electrodes recorded spikes with SNR > 2 out to 12 weeks post-implant. Results from this study are being used to improve the design and specifications of exposed tip lengths of microelectrodes for chronic neural recording in spinal nerves