Assessing Binocular Central Visual Field and Binocular Eye Movements in a Dichoptic Viewing Condition

Podsumowanie

Presented here is a protocol for assessing binocular eye movements and gaze-controlled central visual field screening in participants with central vision loss.

Streszczenie

Macular degeneration typically results in heterogeneous binocular central visual defects. Currently available approaches to assess central visual field, like the microperimetry, can test only one eye at a time. Therefore, they cannot explain how the defects in each eye affect the binocular interaction and real-world function. Dichoptic stimulus presentation with a gaze-controlled system could provide a reliable measure of monocular/binocular visual fields. However, dichoptic stimulus presentation and simultaneous eye-tracking are challenging because optical devices of instruments that present stimulus dichoptically (e.g., haploscope) always interfere with eye-trackers (e.g., infrared video-based eye-trackers). Therefore, the goals were 1) to develop a method for dichoptic stimulus presentation with simultaneous eye-tracking, using 3D-shutter glasses and 3D-ready monitors, that is not affected by interference and 2) to use this method to develop a protocol for assessing central visual field in subjects with central vision loss. The results showed that this setup provides a practical solution for reliably measuring eye-movements in dichoptic viewing condition. In addition, it was also demonstrated that this method can assess gaze-controlled binocular central visual field in subjects with central vision loss.

Wprowadzenie

Macular degeneration is generally a bilateral condition affecting central vision and the pattern of visual loss can be heterogeneous. The central visual loss could be either symmetrical or asymmetrical between two eyes¹. Currently, there are several techniques available to assess the central visual field in macular degeneration. The Amsler grid chart contains a grid pattern that can be used to manually screen central visual field. Automated perimeters (e.g., Humphrey visual field analyzer) present light flashes of varying brightness and sizes in a standardized ganzfeld bowl to probe the visual field. Gaze-contingent microperimetry presents visual stimulus on an LCD display. Micro-perimeters can compensate micro-eye movements by tracking a region of interest on the retina. Micro-perimeters can probe local regions in the central retina for changes in function but can test only one eye at a time. Consequently, micro-perimetric testing cannot explain how the heterogeneous defects in each eye affect the binocular interaction and real-world function. There is an unmet need for a method to reliably assess visual fields in a viewing condition that closely approximates real-world viewing. Such an assessment is necessary to understand how the visual field defect of one eye affects/contributes to the binocular visual field defect. We propose a novel method for assessing central visual field in people with central visual loss under dichoptic viewing condition (i.e., when visual stimuli are independently presented to each of the two eyes).

To measure visual fields reliably, fixation must be maintained at a given locus. Therefore, it is important to combine the eye-tracking and dichoptic presentation for binocular assessment. However, combining these two techniques can be challenging due to interference between the illuminating systems of the eye-tracker (e.g., infrared LEDs) and the optical elements of the dichoptic presenting systems (e.g., mirrors of haploscope or prisms of stereoscopes). Alternative options are to use an eye-tracking technique that does not interfere with the line of sight (e.g., scleral coil technique) or an eye-tracker that is integrated with goggles². Though each method has its own benefits, there are disadvantages. The former method is considered invasive and can cause considerable discomfort³ and the latter methods have low temporal resolutions (60 Hz)⁴. To overcome these issues, Brascamp & Naber (2017)⁵ and Qian & Brascamp (2017)⁶ used a pair of cold mirrors (which transmitted infrared light but reflected 95% of the visible light) and a pair of monitors on either side of the cold mirrors to create a dichoptic presentation. Infrared video-based eye-tracker was used to track eye movements in the haploscope setup^7,8.

However, using a haploscope-type dichoptic presentation has a drawback. The center of rotation of the instrument (haploscope) is different from the center of rotation of the eye. Therefore, additional calculations (as described in Appendix – A of Raveendran (2013)⁹) are required for proper and accurate measurements of eye movements. In addition, the planes of accommodation and vergence must be aligned (i.e., demand for accommodation and vergence must be the same). For example, if the working distance (total optical distance) is 40 cm, then the demand for accommodation and vergence is 2.5 diopters and 2.5-meter angles, respectively. If we align the mirrors perfectly orthogonal, then the haploscope is aligned for distant viewing (i.e., required vergence is zero), but the required accommodation is still 2.5D. Therefore, a pair of convex lenses (+2.50 diopters) must be placed between the eye and mirror arrangement of haploscope to push the plane of accommodation to infinity (i.e., required accommodation is zero). This arrangement necessitates more space between the eye and mirror arrangement of haploscope is required, which takes us back to the difference in centers of rotation. The issue of aligning planes of accommodation and vergence can be minimized by aligning the haploscope to the near viewing such that both the planes are aligned. However, this requires measurement of inter-pupillary distance for every participant and the corresponding alignment of haploscope mirrors/stimulus presenting monitors.

In this paper, we introduce a method to combine infrared video-based eye-tracking and dichoptic stimulus presentation using wireless 3D shutter glasses and 3D-ready monitors. This method does not require any additional calculations and/or assumptions like those used with the haploscopic method. Shutter glasses have been used in conjunction with eye trackers for understanding binocular fusion¹⁰, saccadic adaptation¹¹, and eye-hand coordination¹². However, it should be noted that stereo-shutter glasses used by Maiello and colleagues^10,11,12 were the first-generation shutter glasses, which were connected through a wire to synchronize with the monitor refresh rate. Moreover, the first-generation shutter glasses are commercially unavailable now. Here, we demonstrate the use of commercially available second-generation wireless shutter glasses (Table of Materials) to present dichoptic stimulus and reliably measure monocular and binocular eye-movements. Additionally, we demonstrate a method to assess monocular/binocular visual fields in subjects with central visual field loss. While dichoptic presentation of visual stimulus enables monocular and binocular assessment of visual fields, binocular eye tracking under dichoptic viewing condition facilitates visual fields testing in a gaze-controlled paradigm.

Access restricted. Please log in or start a trial to view this content.

Protokół

All the procedures and protocol described below were reviewed and approved by the institutional review board of Wichita State University, Wichita, Kansas. Informed consent was obtained from all the participants.

1. Participant selection

1. Recruited participants with normal vision (n=5, 4 females, mean ± SE: 39.8 ± 2.6 years), and with central vision loss (n=15, 11 females, 78.3 ± 2.3 years) due to macular degeneration (age-related/juvenile). Note that grossly different ages of the two groups was secondary to demographics of the subjects with central vision loss (age-related macular degeneration affects older subjects and is more prevalent in females). Further, the goal of this study was not comparing the two cohorts.

2. Preparation of the experiment

1. Use a wireless 3D active shutter glasses (Table of Materials) that can be synced with any 3D-ready monitor. For the shutter glasses to be active, there should be no interference between the infrared transmitter (a small pyramid-shaped black box) and the infrared receiver (sensor) on the nose bridge of the shutter glasses.
2. Display all the visual stimuli on a 3D monitor (1920 x 1080 pixels, 144 Hz). For the monitor and the 3D glasses to work seamlessly, ensure that appropriate drivers are installed.
3. Use a table-mounted infrared video-based eye-tracker (Table of Materials) that is capable of measuring eye movements at the sampling of 1000 Hz for this protocol. Separate the infrared illumination and camera of the eye-tracker use any tripod with adjustable height and angle (Table of Materials) to hold them firmly in place. Place the camera at a distance of 20-30 cm from the participant and place the screen at a distance of 100 cm from the participant.
4. Use an infrared reflective patch (Table of Materials) to avoid the interference between infrared illumination of the eye-tracker and the infrared system of the shutter glasses (Figure 1, Right).
5. Use commercially available software (Table of Materials) to integrate shutter glasses and 3D ready monitor for dichoptic presentation of visual stimuli to control the eyetracker.
6. To stabilize the head movements, use a tall and wide chin and forehead rest (Table of Materials) and clamp it to an adjustable table. The wide dimension of the chin and forehead rest allows comfortable positioning of participants with the shutter glasses on.
   NOTE: Figure 1 shows the setup for eye-tracking with dichoptic stimulus presentation using 3D shutter glasses and 3D-ready monitor. The infrared reflective patch was strategically placed below the infrared sensor on the nose bridge of 3D shutter glasses (Figure 1, Right).
7. Minimize the leakage of luminance information by deactivating the light-boost option in the 3D ready monitor. The leakage of luminance information from one eye to the other eye is known as luminance leakage or crosstalk¹³. This is prone to occur with the stereoscopic displays in the high luminance conditions.
8. Because of the shutters, the amount of infrared illumination (from the eye-tracking system) reaching the pupil can be significantly reduced¹³ – on an average, approximately 65% of luminance was reduced (Supplementary Table 1). To overcome this, increase the strength of the infrared LEDs of the eyetracker to 100% or (the maximum setting) from the default power setting. When using the infrared video-based eye-tracker (Table of Materials) change this setting in the “Illumination power” settings in the left bottom screen as shown in Figure 2.

3. Running the experiment

NOTE: The main experiment of this study was binocular eye tracking and screening of the central visual field using dichoptic stimulus. The central visual field screening was comparable to the visual field testing of commercially available instruments (Table of Materials). The physical properties of the visual stimulus such as luminance of the target (~22 cd/m²), luminance of the background (~10 cd/m²), size of the target (Goldmann III – 4 mm²), the visual field grid (Polar 3 grid with 28 points, Figure 3), and stimulus duration (200 ms) were identical to the visual field testing of commercially available instruments. Note that these luminance values were measured through shutter glasses when the shutter was ON (Supplementary Table 1). For the purposes of testing discussed here, the luminance of the stimulus was constant unlike visual field testing where the luminance of the stimulus is altered to obtain a detection threshold. In other words, the experiment employed supra-threshold screening and not thresholding. Therefore, the results of the screening were binary responses (stimuli seen or not seen) and not numerical values.

1. Pre-experiment checks
   1. A couple of minutes before the participant arrives for the testing, ensure that both eye tracker and the host computer (that runs the experiment) is turned on and confirm that the host computer is connected to the eyetracker.
   2. As a rule, confirm the synchronization accuracy (using platform specific commands) of the display before beginning the experiment.
2. Initiating the main experiment
   NOTE: The steps below are very platform specific and is contingent on the script that runs the main experiment. See Supplementary Material that contain the samples of the codes used to design and run the experiment.
   1. Initiate the program (See Supplementary Material - ‘ELScreeningBLR.m’) that runs the main experiment from the appropriate interface. When and if prompted by the program, enter the participant information (such as participant ID, test distance) that is needed to save the output data file in the data folder with a unique filename.
   2. A gray screen with instructions such as “Press Enter to toggle camera; Press C to calibrate, Press V to validate” will appear on the screen. At this stage, adjust the camera of the eye-tracker to align with the participant’s pupil as shown in Figure 2.
3. Eye-tracker calibration and validation
   1. Initiate the calibration of the eye-tracker. Instruct the participants to follow the target by moving the eyes (and not head) and look at the center of the target.
   2. After the successful calibration, initiate the validation. Provide the same instructions as the calibration.
   3. Read the results of the validation step (usually displayed on the screen). Repeat the calibration and validation until “good/fair” (as recommended by the eyetracker manual) result is obtained.
4. Drift correction
   1. Once the calibration and validation of the eye-tracker is done, initiate the drift correction.
   2. Instruct the participants to “look at the central fixation target and hold their eyes as steady as possible”.
      NOTE: After the calibration, validation, and drift correction, the eye-tracking will be initiated simultaneously with the main experiment.
5. Visual field screening
   1. Re-instruct/remind the participant about the task that he/she must do during the experiment. Ask subjects to keep both eyes open during the entire testing.
   2. For this visual field experiment, instruct them to hold the fixation at the central fixation target while responding to “any white light seen” by pressing the “enter” button in the response button (Figure 1, Table of Materials). Instruct them not to move the eyes and search for the new white lights. Also, remind them that the brief white lights can appear at any location on the screen.
      NOTE: During visual field screening, the functioning of shutter glasses can be probed using monocular targets that can be fused to form a complete percept (See Supplementary Figure 2 – catch trials).
   3. Re-iterate the instruction to “hold fixation” several times throughout the experiment to ensure the fixation falls within the desired area.
      NOTE: An audio feedback (like an error tone) can be used to alert loss of fixation (like eyes moved outside a tolerance window). When fixation lapses, reinstruct the participant to fixate only on the cross target. The visual stimuli presentation can be temporarily stopped until the participant brings the fixation back within the tolerance window (like central 2°).
   4. At the end of the visual field experiment, the screen will display the result of the testing highlighting the seen and non-seen locations differently (like for example Figure 6).
6. Saving the data file
   1. All the visual field data (say saved as “. mat” file) and eye-movement data (say saved as “.edf” file) will be saved automatically for post-hoc analysis. However, ensure that the files have been saved before quitting the program/platform running the experiment.

4. Analysis

NOTE: The analysis of eye movement and visual field data can be performed in several ways and is contingent on the software used to run the experiment and data format of eye tracker’s output. The steps below are specific to the setup and the program (See Supplementary Materials).

1. Eye movement analysis (post-hoc)
   NOTE: The saved eye movements data file (EDF) is a highly compressed binary format, and it contains many types of data, including eye movement events, messages, button presses, and gaze position samples.
   1. Convert EDF to ASC-II files using a translator program (EDF2ASC).
   2. Run ‘PipelineEyeMovementAnalysisERI.m’ to initialize eye movement analysis and follow the instructions as noted in the code (See Supplementary Materials for the code script).
   3. Run ‘EM_plots.m’, to extract horizontal and vertical eye positions and to plot as shown in Figure 4 and Figure 5.
      NOTE: Eye movement data can be further analyzed to compute fixation stability, detect microsaccades, etc. However, this is beyond the scope of the current paper.
2. Visual fields
   1. To get the reports of visual field test, run ‘VF_plot.m’.
      NOTE: All datasets pertaining to the visual field experiment such as points seen/not seen will be plotted as a visual field map as shown in Figure 6. If a point was seen, then it will be plotted as “green” filled square, otherwise a red filled square will be plotted. No post-hoc analysis for visual field data will be required.

Access restricted. Please log in or start a trial to view this content.

Wyniki

The representative binocular eye-movement traces of one observer with normal binocular vision during two different viewing conditions is shown (Figure 4). Continuous tracking of eye movements was possible when both eyes viewed the stimulus (Figure 4A), and when the left eye viewed the stimulus with the right eye under an active shutter (Figure 4B). As evident from these traces, the proposed method does not impact the quality of eye-...

Access restricted. Please log in or start a trial to view this content.

Dyskusje

The proposed method of measuring eye movements in dichoptic viewing condition has many potential applications. Assessing binocular visual fields in participants with central vision loss that is demonstrated here is one such application. We used this method to assess binocular visual field in fifteen participants with central vision loss to study how binocular viewing influences the heterogeneous central visual field loss.

The most important step in the protocol is positioning (distance from ey...

Access restricted. Please log in or start a trial to view this content.

Materiały

Name	Company	Catalog Number	Comments
3D monitor	Benq	NA	Approximate Cost (in USD): 500 https://zowie.benq.com/en/product/monitor/xl/xl2720.html
3D shutter glass	NVIDIA	NA	Approximate Cost (in USD): 300 https://www.nvidia.com/object/product-geforce-3d-vision2-wireless-glasses-kit-us.html
Chin/forehead rest	UHCO	NA	Approximate Cost (in USD): 750 https://www.opt.uh.edu/research-at-uhco/uhcotech/headspot/
Eyetracker	SR Research	NA	Approximate Cost (in USD): 27,000 https://www.sr-research.com/eyelink-1000-plus/
IR reflective patch	Tactical	NA	Approximate Cost (in USD): 10 https://www.empiretactical.org/infrared-reflective-patches/tactical-infrared-ir-square-patch-with-velcro-hook-fastener-1-inch-x-1-inch
MATLAB Software	Mathworks	NA	Approximate Cost (in USD): 2150 https://www.mathworks.com/pricing-licensing.html
Numerical Keypad	Amazon	CP001878 (model), B01E8TTWZ2 (ASIN)	Approximate Cost (in USD): 15 https://www.amazon.com/Numeric-Jelly-Comb-Portable-Computer/dp/B01E8TTWZ2
Psychtoolbox - Add on	Freeware	NA	Approximate Cost (in USD): FREE http://psychtoolbox.org/download.html
Tripod (Dekstop)	Manfrotto	MTPIXI-B (model), B00D76RNLS (ASIN)	Approximate Cost (in USD): 30 https://www.amazon.com/dp/B00D76RNLS