Galvanic vestibular testing

Timothy C. Hain, MD•Page last modified: July 31, 2021Return to testing index

Galvanic testing or galvanic vestibular stimulation (GVS) refers to use of a weak electric current passed through the region of the ear to determine whether or not the inner ear is functioning. In animals, galvanic stimulation is thought to stimulate the vestibular nerve, including both afferents from the canals and otoliths (Wardman and Fitzpatrick, 2002). The irregular afferents are thought to be more sensitive to galvanic current than the regular afferents, and inhibitory galvanic current is thought to be able to even "ablate" the irregular fibers (Minor and Goldberg, 1991).

In humans, again eye movements are felt to reflect the "weighted sum of inputs from all vestibular end organs (Vailleau et al, 2011), and act on the "spike trigger site". Practically, one would think that electrical current might also stimulate the hair cells, and also that the extent to which the response comes from the nerve or the hair cell might vary.

Concerning the "weighted sum" idea, in each ear there are 3 canals and 2 otolith organs. The general idea is that the canals and presumably their nerves produce eye movements in the plane of the canals. The otolith organs are circularly polarized, and one would expect that galvanic stimulation of their nerves might not affect them to a very great extent as the vectors would cancel out. The canals are arranged on each side so that excitation of the horizontal canal drives the eye slow component away from the ear. For the vertical canals, excitation of the posterior causes downward slow phases and excitation of the superior upward slow phases. These are opposed vertical components, that should in theory cancel out. The vertical canals produce same-direction torsional components, twisting of the top of the eye away from the excited ear, that should in theory, add. Thus one would expect that galvanic stimulation should cause a horizontal and torsional nystagmus, with little vertical, in normal people. In situations where a part of the nerve is dead -- for example vestibular neuritis, where the top part of the nerve is dead, and bottom still working, one would expect a nystagmus with a larger vertical component due to the unopposed nerve from the remaining posterior canal and saccule.

In humans, recent studies of persons with pure hair cell damage (gentamicin ototoxicity) has suggested that galvanic sensitivity is greatly reduced -- suggesting that the hair cells are required for a galvanic response. (Aw et al, 2008) One would not expect this if only the nerve were needed for galvanic responses. On the other hand, no change in galvanic sensitivity was reported in Chinchillas after gentamicin ototoxicity (Hrvonen et al, 2005). A more recent and large study from Vallieu et al (2011) in humans clearly showed strong responses in gentamicin ototoxicity. Other studies have suggested that galvanic stimulation stimulates the nerve (Dieterich et al, 1999). It is clear that there is disagreement and more study is needed. Our take on this is that GVS stimulates the nerve, or perhaps both hair cells and nerve.


Illustration of method of bilateral galvanic stimulation, from Vallieau et al, 2011


Galvanic stimulation is uncomfortable, and for this reason (at least) it is not popular as a clinical test. The basic equipment needed to elicit a galvanic response is a method of passing current into the head through the ear, and a device that allows one to safely limit the amount of current being applied. Between 2 amd 5 ma of current is generally tolerable. Depending on the resistance of the skin, one might need more or less voltage to get 2-5ma, due to Ohm's law. In other words, voltage might go fairly high.

It is best to use large electrodes, as shown above, to minimize discomfort. As the device for this limits current, a large surface area means that the current density travelling through the skin is less than for the situation where small electrodes are used. In the study above, 4ma current was used. This is about as much current as one can tolerate, and still develop a reasonably strong (i.e. about 10 deg/sec) nystagmus. Devices used for galvanic stimulation are generally battery powered, or be otherwise isolated from the "lines" for safety.

One might speculate that the current density through the skin might vary for a large electrode as shown above, and a more sophisticated design might be to design an electrode and control circuit so that there are patches that are controlled individually. However, why go to all this trouble unless there is a very good reason, and we haven't figured out a very good reason to do GVS as yet.

Output from galvanic stimulation

Eye movements:

Galvanic stimulation in humans, when fixation is allowed, produces a mainly torsional eye movement (Severac et al, 2003; Jahn et al, 2003; Aw et al, 2008). This is caused by the addition of torsional vectors and subtraction of vertical vectors. While one would also expect a substantial horizontal movement, this does not seem to be prominent in humans, but this may be due to lack of recording in darkness. Torsion is difficult to suppress in light or darkness, but horizontal and vertical are easily suppressed with fixation. The relative amounts of torsional eye movement as well as well as the entire amplitude of the response is highly variable (MacDougall et al, 2002). As the main nystagmus output from these studies in humans is torsional, and the amplitude is tiny (about 10 deg/sec), simple ENG or VNG recordings are not especially useful. However, response triggered averaging techniques combined with deviation of the ocular axis away from the front-back axis may be worth considering.

MacDougall et al (2012) reported that the scaling was roughly 0.5 deg/sec per mA for torsion as well as horizontal. Thus this is obviously not a large response.

Vailleau et al (2011) studied galvanic stimulation in a far larger number of patients including 60 patients with no vestibular function to caloric testing including many with gentamicin ototoxicity, another 21 who had acoustic neurinomas prior to surgery, and another 26 after surgical"deafferentation".

Vallieau et al (2011) observed a mainly horizontal nystagmus. It may be that torsion was not easy to record in their video system, and so they ignored it.

They explained the horizontal nystagmus by noting that their study was done in the dark, and suggested that others used fixation, which would have suppressed fixation. This is a not entirely true, as MacDougall et al (2004) recorded nystagmus in both light and darkness. Perhaps there was also an element of not being able to record torsion in the Vallieau study as well. In controls, the nystagmus speed averaged only 4 deg/sec, similar to the Macdougall study in normal subjects. In patients with various lesions, responses of about 2 deg/sec were usually seen (see above), somewhat dependent on the side where the current was applied. So this is a stimulus that produces a weak nystagmus, using a painful method (GVS).

The amount of nystagmus in the Vallieau et al study (2011) was "all over the map" -- anything between 0 and 12 deg/sec, without much difference between subjects with gentamicin ototoxicity (where one would expect no response at all), and normal individuals.

Where there was clear and unequivocal damage to the ear on one side (e.g. surgical deafferentation), the ipsilateral side showed no response (of course).

And so the interesting thing in their study was that gentamicin patients often tested out entirely normal. As gentamicin is a hair cell toxin, one would expect that there would be no response at all from the hair cells, and that all of the response would necessarily come from the nerve. Thus this study might suggest that the response in galvanic stimulation comes mainly from the nerve. This might be helpful in clinical situations where one is attempting to determine if the problem is in the hair cells or nerves, should there ever be a treatment that regrows hair cells.

We would have been interested to know about the differences in vector between patients with vestibular neuritis (where one would expect a larger vertical nystagmus), and patients with "dead" ears from surgery, where one would expect the same vector of nystagmus, but half as much.

Postural deviation:

Galvanic stimulation also causes postural sway, and in fact, vestibular afferents mediating postural reactions may be more sensitive than those mediating ocular responses (Bacsi et al, 2003). Postural sway is slower, it is coupled to vestibular input through many more synapses than eye movement, and is not ameniable to response triggered averaging, but may be easier to record.

To summarize:

Galvanic stimulation in humans elicits very weak eye movements, with mainly a torsion and horizontal vector. Because stimulation is limited to about 5 mA, and the size of the responses are on the order of only 5 deg/sec, it seems unlikely that GVS will be very useful for clinical diagnosis.

While one can imagine ways to get the stimulus closer, or better controlled, given the small amount of eye movement so far, success is not assured.