CI Atlas · Beyond Hearing: The Implant for Tinnitus and the Balance System · Module 11

11Building a Vestibular Implant: The Device

A vestibular implant must do what the labyrinth does: sense head rotation, convert it into a neural code, and deliver that code to the right nerve branch. Each step is an engineering compromise.

FFrom motion sensor to nerve: the signal chain

A vestibular implant has three functional stages: motion sensors, a processor, and implanted electrodes at the vestibular nerve. Micro-machined gyroscopes (MEMS) detect angular velocity of the head, one sensor aligned to each of the three semicircular-canal planes, just as the natural canals are arranged. An external/implanted processor converts measured angular velocity into a modulated train of biphasic current pulses, the rate or amplitude of which encodes how fast the head is turning. Electrodes placed near the ampullary nerve of each canal deliver that pulse train to the surviving primary afferents, substituting for the absent hair-cell drive.[2010][2014]

The vestibular implant mirrors the three semicircular canals with three MEMS gyroscopes, each tuned to one rotational axis — pitch, roll and yaw — so the device captures head rotation in the same three planes the labyrinth uses. The motion processor converts each axis’s angular velocity into a baseline-modulated pulse rate, and three electrodes deliver that signal to the matching ampullary nerve. Matching the three gyroscope axes to the three canal planes is what keeps the artificial signal spatially aligned with the patient’s own reflex geometry. Schematic.

TCoding head movement as a modulated baseline

Healthy canal afferents fire continuously at a resting (tonic) rate of roughly 50-100 spikes/s even when the head is still, and they speed up or slow down with rotation. To mimic this, the implant delivers a constant baseline pulse rate (often a few hundred pulses per second) when the head is motionless, giving the brain a steady 'centre' to read from. Rotation toward the implanted canal increases the pulse rate above baseline; rotation the other way decreases it, encoding direction and speed as up- and down-modulation of the baseline. Establishing and sustaining a baseline the brain can comfortably modulate is itself a core design requirement, because a canal that is silent at rest cannot signal slowing.[2015][2017]

An ampullary electrode is meant to excite only its own nerve, but the charge it injects spreads through perilymph and bone to whatever lies nearby. The other two ampullae, the cochlea and the facial nerve all sit within roughly 1 to 3 mm, so as current rises the field swells and captures them — misaligning the evoked eye movement, adding spurious sound, or twitching the face. Confining the field is the core challenge of vestibular-implant design, addressed with current focusing, electrode placement and careful fitting. Schematic.

CThe engineering problems that make it hard

Current spread is the central obstacle: charge intended for one ampullary nerve leaks to neighbouring canal nerves, to the cochlea (causing hearing changes), and to the nearby facial nerve (causing facial twitching). Spread misaligns the response, so stimulating the horizontal-canal electrode may evoke an eye movement that is partly vertical or torsional rather than purely horizontal. Surgical placement is demanding: electrodes must sit close to each ampullary nerve without destroying residual hearing, and threshold and response can drift over weeks as tissue settles. Designers trade off electrode count, current level and pulse rate to maximise selective, canal-appropriate responses while minimising co-stimulation of cochlea and facial nerve.[2010][2017]

All three research devices converge on the same architecture: a modified cochlear-implant body, gyroscopes sensing head rotation about three axes, and separate ampullary electrodes aimed at each of the three semicircular canals. The Geneva–Maastricht group adapted a MED-EL implant, Johns Hopkins built the dedicated Multichannel Vestibular Prosthesis, and the University of Washington pursued a parallel cochlear-implant–based platform. None is commercially available — every system remains investigational. Schematic.

TThe research devices: who built what

The Geneva/Maastricht group (Guyot, Kingma, Guinand, Perez Fornos) used modified cochlear implants fitted with extra vestibular electrodes, a pragmatic 'cochleo-vestibular' hybrid that reuses proven hardware. The Johns Hopkins group (Della Santina) built a purpose-designed Multichannel Vestibular Prosthesis using three gyroscopes and a microprocessor to emulate all three canals, later commercialised with an industry partner. The University of Washington group (Rubinstein, Phillips) implanted patients (often with intractable Meniere's) and studied longitudinal eye-movement responses to baseline and modulated stimulation. No vestibular implant is yet a routine clinical product; all human work to date is investigational, with small numbers of implanted patients across these centres.[2014][2010][2016]

Case 30.11 · Building a Vestibular Implant

During fitting of a vestibular implant, raising the current on the electrode intended for the right horizontal canal produces a brisk eye movement, but the patient also reports a faint sound and develops a visible twitch at the corner of the mouth.

What single phenomenon best accounts for the sound and the facial twitch?

Self-assessment — Module 113 questions

Question 1

In a vestibular implant, the component that detects head rotation is the:

Question 2

Why must a vestibular implant deliver a tonic baseline pulse rate at rest?

Question 3

The Geneva/Maastricht human vestibular implant work was notable for:

Tracked locally in your browser — see /progress for the dashboard.