Purpose
When an individual looks at a target, that target is imaged on the fovea. It is thus foveal fixation that correlates precisely with gaze direction. Recent research has shown that techniques which effectively track the optical projection of fundus landmarks out from the eye afford a more direct measurement of fixation direction. It has also been shown that landmarks such as the fovea and the optic disc can be detected robustly by measuring the amount of polarization change that the surrounding birefringent nerve fibers cause during double passage of a beam of polarized light through them upon fundus reflection in double-pass systems. Polarized near-infrared light is reflected from the foveal and optic disc areas in bow-tie or propeller patterns of polarization states. These patterns are of constant shape, size, and location, for a given type of polarized light entering the eye, and are therefore detectable, offering the opportunity for eye tracking. The major advantage of this eye-fixation detection and tracking method is that it uses true information coming directly from retinal landmarks, as opposed to existing eye-tracking systems which use reflections from other structures, to identify the direction of foveal gaze.
Methods
A coaxial emitter/detector sensor was developed, consisting of a solid state laser diode, a polarizer, a filter and a photodetector. Several such sensors could be used to interrogate different spots on the retina, thus eliminating the need for scanning systems with moving parts. To test the signal quality obtainable, using just one sensor, the test subject was asked to fixate successively on twelve “glow-in-the-dark” targets located on a circle around the emitter/detector to simulate the retina’s being interrogated by twelve sensors placed on a 3-deg diameter circle surrounding the projection of the fovea (Fig 1).
Results
The resulting signals were plotted (Fig. 2). They are close to the ideal sine wave that would have been recorded from an ideal propeller-type birefringence pattern from a human fovea. We deem these results satisfactory, considering factors such as imperfect bow-tie, imperfect fixation by the test subject, and signal noise.
Conclusions
The method can be used in the detection of central fixation for medical and other purposes. It does not require calibration, strict restrictions on head position, or head-mounted appliances.