Eye tracking as a diagnosis tool for eye movement disorders—a look at nystagmus

This study conducted at the Department of Optometry & Vision Sciences at the University of Melbourne focused on eye movement disorders, in particular nystagmus, and the possibility of using eye tracking as an established form of diagnosis tool.


The study was conducted at the Department of Optometry & Vision Sciences at the University of Melbourne, Australia, by Dr Larry Abel and his team. It focused on eye movement disorders, in particular nystagmus, and the possibility of using eye tracking as an established form of diagnosis tool.

The research lab, which has been doing clinical testing "on the side" for years, used a Tobii Pro TX300 eye tracker to trial and test for this purpose. Successful results would mean being able to provide high-quality nystagmus evaluation and other ocular motor testing, as well as being able to make them available to referring clinicians in a teaching clinic. It would also mean being one step closer to making it a new and regular part of the department's clinical activities, and for trainee optometrists to see eye tracking as a standard and normal clinical tool.

Development of a system that could be used by a wider range of clinicians who do not also have an engineering background could be of great benefit to patients of congenital nystagmus. We will continue our diagnosis with our existing lab while working on the establishment of eye tracking as a clinical diagnosis tool for a broader spectrum of users.

Larry Abel, PhD, Senior lecturer, Department of Optometry and Vision Sciences, The University of Melbourne

Research objectives

The objective of this trial study was to find out whether it is possible to develop a clinical facility for evaluation of nystagmus and other disorders that could eventually be run by less specialized clinicians. The major focus is on nystagmus identification and characterization, but future research will ideally also cover the ability to evaluate saccadic and smooth pursuit eye movements more quantitatively.

The first goal would be for clinicians to be able to conduct tests, with the results being evaluated by specialists, such as the authors of this study. The longer-term goal would be to try to develop more automated means of screening, accessible to more diverse backgrounds of users.

Because good eye movement recordings are essential for proper nystagmus evaluation and for quantitative assessment of saccades eye tracking plays a central role in the used methods and in this study.

Tools and methods

During the trials in the laboratory comparisons were largely made between the Tobii Pro TX300 eye tracker and an old Microguide 1000 limbus tracker. This is the classic analogue technology that dates back to the 1960s whereby optoelectronics is used to detect horizontal motion of the edges of the iris. Some comparisons were also made with a Tobii 1750 Eye Tracker and an Eyelink II from SR Research. The lab also uses an SMI high-speed HED system, but not for these sorts of assessments, given its monocular nature. Clinical examination and history-taking always precede any eye movement recording, and their results often guide the recording session. Eye tracking is then used to confirm, augment or refute initial clinical impressions.

The eye tracking tests consist of patients fixating a range of target positions across their visual field, both horizontally and vertically. Waveforms present are identified and classified as congenital (and if so, which specific waveform) or acquired nystagmus. Quantification of the duration of stable foveation periods (if any) is being carried out when warranted. These foveation periods can be a key factor in determining the potential visual acuity of a patient and can only be observed with eye movement recordings with both good temporal and spatial resolution.

Current laboratory practice for these kinds of tests is to export data from the currently used Tobii 1750 Eye Tracker or to digitize data from the Microguide limbus tracker and enter these into Matlab where they are displayed and quantified. The new Matlab interface of the SDK for the Pro TX300 eye tracker was still under development and not available at the time of the trial so no specific Matlab routines were developed for this study. Instead, the raw data was used and processed in the existing Matlab routines to get the relevant waveforms.

The Tobii TX300 is fast enough to provide very good temporal resolution of waveforms and the Tobii software interface is by far the most user-friendly of all eye tracker manufacturers. Its setup is also extremely user-friendly and implementation of stimuli far easier than other more complex eye tracking software.

Larry Abel, PhD, Senior lecturer, Department of Optometry and Vision Sciences, The University of Melbourne

Eye tracking—a high potential tool for clinical diagnosis

The trial period gave some interesting results.

Only eye tracking recordings can distinguish congenital from acquired nystagmus in cases where clinical presentations are ambiguous. They can also distinguish congenital from latent/manifest latent nystagmus—these related conditions are managed somewhat differently. Waveform quantification can also be used to predict best possible visual outcome. Besides nystagmus, saccadic assessment can be used to identify mild internuclear ophthalmoplegia, a common sign of multiple sclerosis, and to help identify myasthenia gravis.

In some cases, the Pro TX300 eye tracker provided diagnostic waveform information that would have been difficult to obtain otherwise. Furthermore, a real potential advantage of the Pro TX300 is the ease of use of Tobii Pro Studio, so that clinical staff could be trained to do assessments with the evaluation then being carried out by specialists when necessary.

An example graph providing diagnostic waveform information .
An example graph providing diagnostic waveform information .

In the above figures you can see a gaze-evoked nystagmus horizontally (blue traces), with either a linear or decreasing velocity slow-phase component. Notice that when the patient looks back to centre from lateral gaze (arrows), the nystagmus in primary position is now beating the other way. This is called rebound nystagmus and it has never been reported as congenital. Also important in these figures (and unavailable with a limbus tracker) are the vertical eye position traces. You can see here that there is a sustained downbeat nystagmus present no matter what direction this patient's horizontal component is; this means that most of the time the eyes were moving diagonally. This patient was referred to the lab as having congenital nystagmus but the above recording is dissimilar in many ways from this benign condition and as a result further clinical evaluation and close follow-up were recommended.

However, in its present state, several clinically important assessments for nystagmus testing were either difficult or impossible. These included:

  • Monocular calibration with binocular recording (i.e., to be able to block the view of one eye but still record its behavior). Once recording has started, the ability to alternate occlusion and observe the effect.*
  • Operator selection of target location and timing
  • Real-time monitoring of eye position
  • A means of adding target location and manual marker triggering to the eye position data.

*) Since this study was conducted, the Tobii Calibration and Verification tool has become available. Tobii Calibration and Verification tool is an application that measures and reports angles of deviation in the participant's gaze patterns for screening of eye alignment and diagnostics of ophthalmic disorders, such as nystagmus, strabismus (crossed eyes), or amblyopia (lazy eye). It enables measurement of eye movements for instance during the rehabilitation progress of a participant's therapy. The tool consists of software (running on Windows) and an occluder that transmits Infrared light.

Looking forward

The initial goal now is to make a system that most clinicians can operate but will still need input from a specialist to evaluate the results. The longer-term goal would be to develop a neural network-based expert system that could autonomously do waveform classification and, if possible, quantification. This would parallel work in electrodiagnostics and could make such testing far more widely available. Implementation would probably require recruitment of a PhD student with an electrical engineering/computer science background and expertise in signal processing.

Use our contact form

Contact Sales