Exploring eye movements in patients with glaucoma when viewing a driving scene

Abstract

Background: Glaucoma is a progressive eye disease and a leading cause of visual disability. Automated assessment of the visual field determines the different stages in the disease process: it would be desirable to link these measurements taken in the clinic with patient's actual function, or establish if patients compensate for their restricted field of view when performing everyday tasks. Hence, this study investigated eye movements in glaucomatous patients when viewing driving scenes in a hazard perception test (HPT).

Methodology/principal findings: The HPT is a component of the UK driving licence test consisting of a series of short film clips of various traffic scenes viewed from the driver's perspective each containing hazardous situations that require the camera car to change direction or slow down. Data from nine glaucomatous patients with binocular visual field defects and ten age-matched control subjects were considered (all experienced drivers). Each subject viewed 26 different films with eye movements simultaneously monitored by an eye tracker. Computer software was purpose written to pre-process the data, co-register it to the film clips and to quantify eye movements and point-of-regard (using a dynamic bivariate contour ellipse analysis). On average, and across all HPT films, patients exhibited different eye movement characteristics to controls making, for example, significantly more saccades (P<0.001; 95% confidence interval for mean increase: 9.2 to 22.4%). Whilst the average region of 'point-of-regard' of the patients did not differ significantly from the controls, there were revealing cases where patients failed to see a hazard in relation to their binocular visual field defect.

Conclusions/significance: Characteristics of eye movement patterns in patients with bilateral glaucoma can differ significantly from age-matched controls when viewing a traffic scene. Further studies of eye movements made by glaucomatous patients could provide useful information about the definition of the visual field component required for fitness to drive.

Figure 1. Eye movements for patients and controls.

Graphs showing the average response in each eye movement parameter for the patients and the controls with (a) average number of saccades per second; (b) average number fixations per second; (c) number of smooth pursuits per second of film; (d) average saccade amplitude (size in degrees); (e) average fixation duration (ms); and (f) average ‘smooth pursuit’ duration (ms). Each symbol represents the results from one of the 26 films. If there were no differences in the averages then symbols would fall exactly on the line of unity. For example, (a) indicates that the patients made, on average, more saccades across the films than the control subjects, which illustrates the statistically significant effect found in the GLM ANOVA.

Figure 2. Novel ‘Point of Regard’ Analysis.

Examples of the BCE ‘point of regard’ analysis superimposed on frames from some of the HPT films. The centre of the red ellipse represents the mean ‘point of regard’ of the controls with the spatial extent of the ellipse being one standard deviation from this centre along two principal axes, theoretically affording ‘coverage’ of approximately 67% of the ‘point of regard’ locations in a given frame. Note how the location and spatial extent of this ellipse changes with each frame of the film as the control subjects' view is drawn to different aspects of the changing road scene. For each frame the number of patients' points of regard (blue symbols) ‘captured’ by the ellipse can be automatically counted and compared to the expected value under the null hypothesis that the average ‘viewing area’ is the same in the two groups. Note that not all nine patients are recorded in each frame because of blinks, loss of data, and variable responses. Also, note that in the UK vehicles are driven on the left hand side of the road. See Audio/Video - S1.

Figure 3. Visual fields, dynamic ‘point of regard’ and HPT films – example.

HFA grayscales for monocular (30-2) and binocular integrated visual fields for one of the participants with bilateral glaucoma. The integrated visual field is scaled and superimposed on the point of regard HPT film to give a representation of the restricted field of view. The panels show a montage of selected frames from one of the films. In the top frame, whilst most of visually healthy subjects (red symbols) fixate on the pedestrian with a pram/pushchair, seeing them as a potential hazard, the patient (blue symbol) fails to alter their gaze. Similarly in the next frames most of control subjects quickly move their fixation towards the looming hazard of the car hazardously turning on to the main street from the right intersection. Note that in the UK vehicles are driven on the left side of the road. The patient's view of this hazard is masked by their defect and they do not alter their gaze even at the late stage where the camera car ‘brakes’ to take evasive action (bottom frame). This patient has a binocular visual field that is on the borderline of satisfying the UK guideline for visual field fitness to drive. This montage is shown as a film in Audio/Video - S2 and Audio/Video - S3.