Publications / Arvo Abstracts /

Pupil-tracking-based determination of visual acuity

Purpose: The measurement of visual acuity (VA) is an essential tool in daily clinical practice. Standard procedures are based on question-and-answer protocols involving a conscious response from the patient, which is slow and can be difficult or even unfeasible in some cases. The purpose of this work is to develop and test a procedure to measure VA based on subject’s fixation changes, which are fairly instinctive and require a lower level of consciousness.

Methods: We developed a custom-made pupil tracking system and software to detect subject’s fixation changes in response to stimuli presentation. VA was measured in 5 subjects with a two-alternative (left/right) forced-choice-like protocol using pupil tracking data and, simultaneously, using subject’s answers. Stimuli consisted of 0.5-deg-wide maximum-contrast checkerboards 0.5 deg off-center. The experiment consisted of a randomized sequence of 50 2-sec presentations comprising 10 repetitions for 5 square sizes in 0.1 steps of equivalent decimal VA around the smallest square size still rendering the checkerboard visible, which was determined by adjustment prior to the experiment. Between presentations, a central fixation target was turned on for 2 sec. Subjective and pupil-tracking VA estimates were obtained by sigmoid fitting the percentage of correct answers and correct average direction of fixation during stimulus presentation, respectively.

Results: The correlation between measured VAs with both methods was very good (R=0.988) with virtually unit slope (1.013). Differences between VA estimates were small but positive in all 5 subjects (0.13 to 0.04 range; 0.07 mean), implying a slight underestimation by the pupil-tracking method, possibly due to the simple decision criterion used to determine whether the subject was fixating on the stimulus (i.e., seeing it).

Conclusions: We developed a pupil tracking-based method reliably providing VA estimates similar to those obtained by a standard subjective method, suppressing the requirement of a subject’s decision and answer. This method can be of interest in patients with difficulty to understand and/or perform the subjective task, and, with an optimized protocol, could speed up VA measurement in clinical settings.

This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 – May 2, 2019.

If you like it, please share it...Tweet about this on Twitter0Share on Facebook0