Using eye movements to detect visual field loss: a pragmatic assessment using simulated scotoma

Abstract

Glaucoma is a leading cause of irreversible sight-loss and has been shown to affect natural eyemovements. These changes may provide a cheap and easy-to-obtain biomarker for improving disease detection. Here, we investigated whether these changes are large enough to be clinically useful. We used a gaze-contingent simulated visual field (VF) loss paradigm, in which participants experienced a variable magnitude of simulated VF loss based on longitudinal data from a real glaucoma patient (thereby controlling for other variables, such as age and general health). Fifty-five young participants with healthy vision were asked to view two short videos and three pictures, either with: (1) no VF loss, (2) moderate VF loss, or (3) advanced VF loss. Eye-movements were recorded using a remote eye tracker. Key eye-movement parameters were computed, including saccade amplitude, the spread of saccade endpoints (bivariate contour ellipse area), location of saccade landing positions, and similarity of fixations locations among participants (quantified using kernel density estimation). The simulated VF loss caused some statistically significant effects in the eye movement parameters. Yet, these effects were not capable of consistently identifying simulated VF loss, despite it being of a magnitude likely easily detectable by standard automated perimetry. Glaucoma is a chronic eye disease affecting 1 in 28 people aged 40-80 years 1. It is characterised by progressive, irreversible visual field (VF) loss. Early detection is therefore crucial 2. Currently, however, as many as 50% of cases remain undiagnosed 3-5 , or are diagnosed late: after substantial vision loss has already occurred 6. At present, successful detection of glaucoma requires a detailed assessment by a specialist clinician, including measurements of intraocular pressure (IOP), VF loss by standard automated perimetry, and inspection of the optic nerve head. Unfortunately, many adults do not attend routine eye-checks due to associated costs (real and perceived), lack of awareness, aversion to the methods used, and lack of understanding about their purpose 7,8. One way to improve glaucoma detection would be to perform proactive screening. However, this is not economical using traditional technologies, owing to the high cost of the personnel and equipment involved 9. What is needed, therefore, is an inexpensive, automated screening tool to identify high-risk individuals. Modern eye tracking technologies may be able to provide such a solution. Previous studies have shown that natural eye movements are altered in instances of glaucomatous VF loss. For example, our research has demonstrated that glaucoma patients exhibit differences in saccade frequency and gaze spread when free-viewing images, relative to normally sighted controls 10. More recently, we found similar differences in eye-movements between the two eyes of glaucoma patients with asymmetric VF loss, when free-viewing pictures monocularly 11. Several other studies have also likewise confirmed a link between simulated ...

Does the sampling frequency of an eye tracker affect the detection of glaucomatous visual field loss?

“…It is well established that eye movements are affected by either real or simulated glaucomatous visual field defects. [9][10][11][12][22][23][24][25] However, previous studies were conducted at different sampling frequencies, and there is no evidence to suggest which sampling frequency and eye movement parameters are best suited for screening glaucomatous VFL during a free-viewing task. This study aimed to demonstrate how the sampling frequency of eye trackers influenced eye movement parameters in distinguishing glaucomatous visual field defects from those in healthy eyes.…”

“…However, they did observe smaller saccadic amplitudes and a restricted BCEA in the worse eye compared with the better eye. In a follow-up study, Asfaw et al 10 assessed eye movements in simulated VFL of varying degrees based on real glaucomatous VFL. They noted that saccadic amplitude and BCEA were reduced in the advanced VFL group compared with the no VFL group.…”

How Free-Viewing Eye Movements Can Be Used to Detect the Presence of Visual Field Defects in Glaucoma Patients

“…Asfaw et al ( 24 , 25 ) showed that the same patient displays differences in viewing behavior depending on with which eye they performed a free viewing task. The spread of fixation locations was smaller in the worse eye of glaucoma patients compared to the better eye ( 24 ).…”

Different Eye Movement Behaviors Related to Artificial Visual Field Defects–A Pilot Study of Video-Based Perimetry

“…A set of features was extracted by kernel principal component analysis on a saccade map, and these features were used as the input of a Naïve Bayes classifier that separated the patients from the controls. Asfaw et al [37] simulated artificial VFDs from a real glaucoma patient and performed a free-viewing task using both images and videos. They evaluated the clinical usefulness of four eye-movement parameters by plotting receiver operating characteristics curves.…”