Multichannel vestibular prosthesis employing modulation of pulse rate and current with alignment precompensation elicits improved VOR performance in monkeys

Abstract

An implantable prosthesis that stimulates vestibular nerve branches to restore the sensation of head rotation and the three-dimensional (3D) vestibular ocular reflex (VOR) could benefit individuals disabled by bilateral loss of vestibular sensation. Our group has developed a vestibular prosthesis that partly restores normal function in animals by delivering biphasic current pulses via electrodes implanted in semicircular canals. Despite otherwise promising results, this approach has been limited by insufficient velocity of VOR response to head movements that should inhibit the implanted labyrinth and by misalignment between direction of head motion and prosthetically elicited VOR. We report that significantly larger VOR eye velocities in the inhibitory direction can be elicited by adapting a monkey to elevated baseline stimulation rate and current prior to stimulus modulation and then concurrently modulating ("co-modulating") both rate and current below baseline levels to encode inhibitory angular head velocity. Co-modulation of pulse rate and current amplitude above baseline can also elicit larger VOR eye responses in the excitatory direction than do either pulse rate modulation or current modulation alone. Combining these stimulation strategies with a precompensatory 3D coordinate transformation improves alignment and magnitude of evoked VOR eye responses. By demonstrating that a combination of co-modulation and precompensatory transformation strategies achieves a robust VOR response in all directions with significantly improved alignment in an animal model that closely resembles humans with vestibular loss, these findings provide a solid preclinical foundation for application of vestibular stimulation in humans.

Figures

FIG. 1

A SCC coordinate system used for description of head rotation and 3D VOR eye rotation responses. +X, +Y, and +Z vectors intersect at the stereotactic origin (i.e., the midpoint of the interaural axis) and are perpendicular to the stereotactic coronal, sagittal, and horizontal planes, respectively (Della Santina et al. 2005b). Left-anterior/right-posterior (+LARP), right-anterior/left-posterior (+RALP), and +Z vectors represent the cardinal axes of the SCC coordinate system. Curved arrows depict the direction of positive rotation about each axis. B Thirty-eight unique vectors were selected to characterize the VOR responses to prosthetic stimulation. Each vector corresponds to a unique head rotation and desired eye response movement and contains three elements representing the stimulus intensity on each electrode (targeting the horizontal, superior, and posterior canal ampullary nerves). The direction of each vector represents the desired axis of eye rotation, and the length represents the magnitude of the desired eye velocity resulting from each . Ideally, eye rotations exclusively about the Z, LARP, or RALP axes would be elicited by electrical stimulation of only the left horizontal, left-anterior, or left-posterior canal electrodes, respectively.

FIG. 2

A–C Examples of cycle-averaged eye movements for animal 1’s left horizontal semicircular canal in response to stimulation steps from a zero baseline (0 pps, 0 μA,) to (600 pps, 107 μA) (A), (200 pps, 150 μA) (B), and (600 pps, 150 μA) (C). All pulses were 150 μs/phase, biphasic, symmetric, and charge balanced. Eye responses are displayed as mean (thick trace) ± SD (thin traces) for eye rotational velocity components about Z (red, solid trace), LARP (green, dashed trace), or RALP (blue, dot–dashed trace) axes. Dashed lines indicate stimulus onset. D–I Surface plots of peak eye velocities of responses to 36 unique stimulation steps in pulse rate and pulse current amplitude from a zero baseline (0 pps, 0 μA). Three of these responses are depicted in panels A, B, and C and correspond to the triangle, circle, and diamond in panel D, respectively. Peak eye response velocities increase with rate and with current amplitude. Responses are shown for all three tested semicircular canals for both tested monkeys.

FIG. 3

Cycle-averaged eye response traces following adaptation of animal 1’s left horizontal SCC ampullary nerve to baseline stimulation rate and current of 94 pps, 107 μA (A–D) or 200 pps, 121 μA (E–H). Eye responses are displayed as mean (thick trace) ± SD (thin traces) for eye rotational velocity components about Z (red, solid trace), LARP (green, dashed trace), or RALP (blue, dot–dashed trace) axes. The dashed line indicates stimulus onset time. A, E Step in pulse rate to 600 pps (maximum delivered pulse rate) while current was held constant. B, F Step in current to 150 μA (maximum delivered current amplitude) while pulse rate was held constant. C, G Step in both pulse rate and current up to 600 pps, 150 μA. D, H Step in both pulse rate and current down to (0 pps, 0 μA). Rate modulation down to 0 pps and current modulation down to 0 μA are mathematically identical to co-modulation down to 0 pps, 0 μA, since no stimulation can be delivered when either the pulse rate or current amplitude is 0. Compared to either rate-only or current-only modulation, co-modulation yielded largest excitatory and inhibitory eye movements regardless of baseline. Considering the goal of effectively encoding both excitatory and inhibitory head rotations, co-modulation from a high baseline yielded the best overall performance. Although responses to excitatory stimuli in this condition were less robust than observed for the low-baseline paradigm, responses for inhibitory stimuli (i.e., down-modulating both pulse rate and current amplitude from a high baseline) were more robust than those achieved with the low-baseline paradigm, with responses approaching 3,000 °/s2. In some cases, stimulus duration and intensity were sufficient that the resulting slow-phase VOR eye movement drove the eye to near its range of motion limit shortly after modulation onset. As a result, the downward deflections seen in A, C, and G represent nystagmus quick phases.

FIG. 4

Peak eye velocities from each of six tested SCCs in response to co-modulation above (excitatory) or below (inhibitory) either low or high baseline stimulation that the monkey was adapted to. The trends observed in Fig. 3 for one SCC were consistent for all six. At high baseline, excitatory velocity was smaller but inhibitory velocity was larger. * represents significant difference between two datasets at the p < 0.01 level (paired t test).

FIG. 5

A, B Cycle-averaged eye velocity traces for both monkeys with mean (thick trace) and SD (thin traces) in response to uncompensated and precompensated stimuli (with eight transformation matrices). The dashed line indicates stimulus onset time. The intended axis of rotation was Z, LARP, or RALP. In all cases, precompensated responses were better aligned than uncompensated responses to intended axis of rotation and closer to desired eye velocity (monkey 1, 81 °/s; monkey 2, 28 °/s). SD tended to be larger near the latter portions of the response traces due to recovery quick phase eye movements that were more likely to occur at those times and were not synchronized with the stimulus onset. C, D Response vectors are depicted on a 3D sphere for easier visualization. The thin arrow represents intended axis of rotation, the dashed line represents uncompensated response vector, and the solid line represents precompensated response vector

FIG. 6

Misalignment and eye velocity in response to all 38 uncompensated and precompensated (with eight transformation matrices) stimulation vectors for both monkeys. The X-axis represents azimuth (equivalent to longitude), and the Y-axis represents elevation (equivalent to latitude) on a theoretical sphere of possible eye rotations. Projections of each data point are depicted on the left wall of each panel for easier visual comparison. On average, misalignment values were smaller, and eye velocities were closer to the desired velocity with less variance for responses to precompensated stimuli than for responses to uncompensated stimuli. Dashed lines represent desired misalignment (0 ° in all cases) or desired eye velocity. The square, circle, and triangle depict vectors with intended axes of Z, LARP, and RALP, respectively