Videonystagmography

By Marcello Cherchi, MD PhD

For patients

Here are answers to some of the most common questions about videonystagmography testing.

What is this test?	Videonystagmography (VNG) quantifies certain eye movements.
What is this test looking for?	This test looks for certain disorders of eye movements that are sometimes indicative of a balance dirorder.
Is this test experimental or investigational?	This test is approved by the Food and Drug Administration.  It is neither experimental nor investigational.
What happens during this test?	During this test the audiologist or technician will have you wear goggles containing a small camera that looks at your eye movements while you perform various visual tasks.
Is this test uncomfortable?	The test is not uncomfortable.
How long does this test take?	Depending on the context, the test can take 45 – 60 minutes.
Do I have to prepare for this test?	It is best to avoid wearing eye makeup.  If possible, try to avoid taking medications such as meclizine (Bonine®) or dimenhydrinate (Dramamine®) for 24 hours prior to this test.
Are there any special instructions for what to do after the test?	There are no special instructions for what to do after this test.

For clinicians

Overview

Videonystagmography (VNG) is a test that can quantify horizontal and vertical eye movements; some platforms claim to identify torsional eye movements as well. Equipment for performing VNG is more expensive than for electronystagmography (ENG), but VNG is easier for an audiologist or ototechnician, and is less subject to certain errors and calibration problems, though it also has limitations. VNG can be helpful in identifying eye movements that are diagnostic of a variety of vestibular and neurological disorders.

Introduction

Videonystagmography (VNG) uses digitized data from an infrared video camera placed over the eye and applies image recognition algorithms to each individual frame of the infrared video of eye movements. These algorithms attempt to identify the pupil and locate its center, whose horizontal and vertical coordinates are then plotted over time to generate a tracing of the eye movements.

Identifying the pupil from an infrared image of the eye, and calculating its centroid, is not always straightforward, and a variety of algorithms have been applied to this computational problem (Min-Allah et al. 2021; Romaguera et al. 2021), each with its strengths and weaknesses.

Advantages of VNG include that it is technically easier than electronystagmography (ENG); calibration is relatively simple and usually only needs to be performed once during a study.

Disadvantages include that the equipment is more expensive than ENG; the results are subject to artifact from factors (such as eye makeup or irregular pupils) that confound image recognition algorithms; and it cannot detect eye movements when the eye is closed (and thus cannot detect certain pathologies (Mulch and Bonner 1976)). Although some literature describes algorithms that process ocular torsion (Otero-Millan et al. 2015) these are not yet in wide use.

There are entire textbooks written on the subject of videonystagmography (McCaslin 2013).

Equipment needed

Videonystagmography is performed by an audiologist or ototechnician. The test requires a specialized set of goggles with one or two infrared cameras (depending on whether a monocular or binocular recording is being performed), and a method for occluding the patient’s vision in one or both eyes (to facilitate parts of the test that require the patient to be in darkness). If caloric testing will be performed, then an irrigator is needed.

The Figure below, from Halmagyi and colleagues (Halmagyi et al. 2020), shows a typical VNG setup with infrared Frenzel goggles.

How to perform the test

While there is no standardized set of subtests for oculomotor testing, most clinical vestibular laboratories include spontaneous nystagmus, eccentric gaze, horizontal and vertical saccades, horizontal and vertical smooth pursuit, upright positional testing (sometimes referred to as vertebral artery testing), the Dix-Hallpike maneuver and horizontal optokinetic testing. This is a reasonable selection of subtests for detecting a fairly broad range of disorders. Depending on the clinical scenario, caloric testing may be included.

What this test assesses

VNG quantifies ocular movements over time while the patient is engaged in various tasks instructed by the audiologist or ototechnician. Most commercially available VNG systems only quantify horizontal and vertical eye movements.

How to interpret the test results

Output from VNG is generally a tracing in which the X‑axis is time, and the Y‑axis is degrees of eye displacement from primary position of gaze; in some systems the horizontal eye movement tracing is depicted in one color, while the vertical eye movement tracing is displayed in another color. In some software packages that claim to track torsional eye movements, this is displayed in a third tracing.

Limitations

Disadvantages of VNG include that the equipment is more expensive than ENG; the results are subject to artifact from factors (such as eye makeup) that confound image recognition algorithms; and it cannot detect eye movements when the eye is closed (and thus cannot detect certain pathologies (Mulch and Bonner 1976)). Although some literature describes algorithms that process ocular torsion (Otero-Millan et al. 2015) these are not yet in wide use.

Contraindications

There are no firm contraindications to VNG.

Pitfalls

VNG employs image recognition algorithms applied to each individual frame of an infrared video of eye movements. Several factors (such as eye makeup and irregular pupils) can interfere with these algorithms, giving rise to multiple potential sources of error in VNG.

Calibration errors

If calibration is performed incorrectly, then the results may be corrupted.

Errors due to the method of pupil detection

One class of errors in VNG arises from the method of pupil detection.

First, an irregularly shaped pupil may be incorrectly processed by an algorithm, and either not be recognized, or its center may be incorrectly located.

Second, since identification of the pupil relies on its being much darker than surrounding tissues, introduction of other dark areas (typically eye makeup) can interfere with correct detection of the pupil, resulting in error.

Third, eye blinks obscure the pupil partially or completely. The algorithms are proprietary and therefore not publicly available, but various algorithms appear to handle this problem in different ways. Some software packages appear to make no effort to detect this, and the resulting tracing shows wild variation. Other software packages appear to plot the centroid of the pupil; during a blink as the top eyelid descends it will cover the upper part of the pupil first, thus even if the pupil is remaining stationary, its calculated centroid will descend; this usually results in a sawtooth looking pattern on the tracings. Yet other software packages, when the pupil cannot be detected, appear to impute a value to the location of the pupil (carried over from the most recent frame in which the pupil was detected) and perpetuate it until the pupil can be successfully identified again in a subsequent frame; this usually results in square wave looking patterns on the tracings.

Errors from goggle slippage

Another class of error in VNG arises from goggle slippage, which results in the camera being displaced relative to the eye. This is generally not significant, but if the goggles are too loose, or if a patient is uncomfortable or very fidgety and tries to adjust the goggles herself or himself and the examiner does not notice, then it can become problematic. On VNG, a translation of 1 mm (of the camera relative to the eye) introduces approximately 5 degrees of error in the corresponding tracing.(Eggert 2007; Gananca et al. 2010) If the examiner becomes aware that this has occurred, then he or she should recalibrate.

Summary of potential sources of error

Overall, the computer algorithms that analyze eye movement videos, generate corresponding tracings and recognize patterns in those tracings are fairly sophisticated, but they are not yet an adequate substitute for reviewing the tracings yourself, or (better) reviewing the original video of the eye movement.

When is the test indicated

For patients with the symptom of disequilibrium it is reasonable to check videonystagmography (VNG).

Diseases that may be diagnosed by this test

Results from videonystagmography may be helpful in diagnosing a number of vestibular and neurological disorders.

References

Eggert T (2007) Eye movement recordings: methods. Dev Ophthalmol 40: 15-34. doi: 10.1159/000100347

Gananca MM, Caovilla HH, Gananca FF (2010) Electronystagmography versus videonystagmography. Braz J Otorhinolaryngol 76: 399-403.

Halmagyi GM, McGarvie LA, Strupp M (2020) Nystagmus goggles: how to use them, what you find and what it means. Pract Neurol 20: 446-450. doi: 10.1136/practneurol-2020-002513

McCaslin DL (2013) Electronystagmography/videonystagmography. Plural Pub., San Diego

Min-Allah N, Jan F, Alrashed S (2021) Pupil detection schemes in human eye: a review. Multimedia Systems 27: 753-777. doi: 10.1007/s00530-021-00806-5

Mulch G, Bonner C (1976) The effect of eye closure upon the pathological vestibular spontaneous nystagmus. Acta Otolaryngol 81: 376-85.

Otero-Millan J, Roberts DC, Lasker A, Zee DS, Kheradmand A (2015) Knowing what the brain is seeing in three dimensions: A novel, noninvasive, sensitive, accurate, and low-noise technique for measuring ocular torsion. J Vis 15: 11. doi: 10.1167/15.14.11

Romaguera TV, Romaguera LV, Piñol DC, Seisdedos CRV (2021) Pupil center detection approaches: a comparative analysis. Computación y Sistemas 25: 67-81. doi: 10.13053/CyS-25-1-3385