Using natural viewing behavior to screen for and reconstruct visual field defects

Digital health and wearable devices for retinal disease monitoring

Digital health is wielding a growing influence across all areas of healthcare, encompassing various facets such as telemedicine, artificial intelligence (AI), and electronic healthcare records. In Ophthalmology, digital health innovations can be broadly divided into four categories: (i) self-monitoring home devices and apps, (ii) virtual and augmented reality visual aids, (iii) AI software, and (iv) wearables. Wearable devices can work in the background, collecting large amounts of objective data while we do our day-to-day activities, which may be ecologically more valid and meaningful to patients than that acquired in traditional hospital settings. They can be a watch, wristband, piece of clothing, glasses, cane, smartphone in our pocket, earphones, or any other device with a sensor that we carry with us. Focusing on retinal diseases, a key challenge in developing novel therapeutics has been to prove a meaningful benefit in patients' lives and the creation of objective patient-centred endpoints in clinical trials. In this review, we will discuss wearable devices collecting different aspects of visual behaviour, visual field, central vision, and functional vision, as well as their potential implementation as outcome measures in research/clinical trial settings. The healthcare landscape is facing a paradigm shift. Clinicians have a key role of collaborating with the development and fine-tuning of digital health innovations, as well as identifying opportunities where they can be leveraged to enhance our understanding of retinal diseases and improve patient outcomes.

Mapping simulated visual field defects with movie-viewing pupil perimetry

PURPOSE

Assessing the quality of the visual field is important for the diagnosis of ophthalmic and neurological diseases and, consequently, for rehabilitation. Visual field defects (VFDs) are typically assessed using standard automated perimetry (SAP). However, SAP requires participants to understand instructions, maintain fixation and sustained attention, and provide overt responses. These aspects make SAP less suitable for very young or cognitively impaired populations. Here we investigate the feasibility of a new and less demanding form of perimetry. This method assesses visual sensitivity based on pupil responses while performing the perhaps simplest task imaginable: watching movies.

METHOD

We analyzed an existing dataset, with healthy participants (n = 70) freely watching movies with or without gaze-contingent simulated VFDs, either hemianopia (left- or right-sided) or glaucoma (large nasal arc, small nasal arc, and tunnel vision). Meanwhile, their gaze and pupil size were recorded. Using a recently published toolbox (Open-DPSM), we modeled the relative contribution of visual events to the pupil responses to indicate relative visual sensitivity across the visual field and to dissociate between conditions with and without simulated VFDs.

RESULT

Conditions with and without simulated VFDs could be dissociated, with an AUC ranging from 0.85 to 0.97, depending on the specific simulated VFD condition. In addition, the dissociation was better when including more movies in the modeling but the model with as few movies as 10 movies was sufficient for a good classification (AUC ranging from 0.84 to 0.96).

CONCLUSION

Movie-viewing pupil perimetry is promising in providing complementary information for the diagnosis of VFDs, especially for those who are unable to perform conventional perimetry.

Characterization of Gaze Metrics and Visual Search Pattern Among Glaucoma Patients While Viewing Real-World Images

PRCIS

We quantified and compared the gaze metrics during target-oriented visual search tasks between glaucoma and healthy controls. On the basis of a mathematical concept we showed that due to glaucoma, focal search becomes prominent over global search.

PURPOSE

Visual search (VS) which is essential for target identification and navigation is significantly impacted by glaucoma. VS metrics can be influenced by differences in cultural exposure or coping strategies, leading to varying VS patterns. This study aimed to explore and label the pattern of VS based on gaze metrics quantified using eye-tracking technology.

METHODS

Twenty-seven glaucoma subjects and 30 healthy controls [median age 51 (14) and 54 (19) y, respectively] underwent a VS experiment during which they had to identify specific targets from real-world images. Eye movements were recorded using a remote eye-tracker and gaze metrics-fixation count (FC), fixation duration (FD), saccade amplitude (SA), and VS time (VST) were computed and compared between the study groups. A Z -score-based coefficient " K " was derived to label the search patterns as global ( K ≤ - 0.1: short FD with long SA), focal ( K ≥+0.1: long FD with short SA), or a combination ( K between ±0.1).

RESULTS

Similar to other ethnicities, Indian glaucoma subjects also exhibited statistically significantly increased FC, FD, and VST ( P =0.01). Healthy controls presented a comparable proportion of focal (47%) and global (42%) search patterns while glaucoma subjects exhibited predominantly focal (56%) than global search patterns (26%, P =0.008).

CONCLUSIONS

This study suggests that glaucoma subjects perform more focal searches during active gaze scanning. This change in viewing behavior reflects underlying compensatory strategies adapted for coping with their visual impairments. These search patterns can be influenced by factors such as saliency which requires further investigation.

Inhibitory control and working memory using saccadic eye movements in primary glaucoma

Glaucoma is a progressive optic neuropathic disorder that significantly impacts the activities of daily life (ADLs) of individuals. Emerging studies indicate degenerative changes in cortical and subcortical regions in individuals with glaucoma, which are associated with cognitive processes and oculomotor control. Cognitive processes involving top-down processes such as attention, planning, task management and execution, are crucial for meeting the demands of everyday tasks, and are affected in glaucoma. This study investigated the executive functions, specifically inhibitory control (IC) and working memory (WM), in individuals with glaucoma compared to age-matched controls, using eye movements. This was achieved through four tasks sensitive to executive functions, including antisaccade, memory-guided prosaccade and antisaccade, and the Go-NoGo tasks. Saccadic eye-movement parameters were also assessed in a prosaccade task, considered as a control condition with minimal IC and WM load. The results indicated that glaucoma is associated with changes in both IC and WM. Increased anticipatory saccadic errors might be linked to inhibitory deficiencies during the preparatory stage of the saccadic suppression mechanism. The increased omission errors in the antisaccade task might be due to the lack of regulation of the WM component. Taken together, these findings provide evidence for the involvement of cognitive deficits in individuals with glaucoma.

Open-DPSM: An open-source toolkit for modeling pupil size changes to dynamic visual inputs

Pupil size change is a widely adopted, sensitive indicator for sensory and cognitive processes. However, the interpretation of these changes is complicated by the influence of multiple low-level effects, such as brightness or contrast changes, posing challenges to applying pupillometry outside of extremely controlled settings. Building on and extending previous models, we here introduce Open Dynamic Pupil Size Modeling (Open-DPSM), an open-source toolkit to model pupil size changes to dynamically changing visual inputs using a convolution approach. Open-DPSM incorporates three key steps: (1) Modeling pupillary responses to both luminance and contrast changes; (2) Weighing of the distinct contributions of visual events across the visual field on pupil size change; and (3) Incorporating gaze-contingent visual event extraction and modeling. These steps improve the prediction of pupil size changes beyond the here-evaluated benchmarks. Open-DPSM provides Python functions, as well as a graphical user interface (GUI), enabling the extension of its applications to versatile scenarios and adaptations to individualized needs. By obtaining a predicted pupil trace using video and eye-tracking data, users can mitigate the effects of low-level features by subtracting the predicted trace or assess the efficacy of the low-level feature manipulations a priori by comparing estimated traces across conditions.

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

PURPOSE

Evidence suggests that eye movements have potential as a tool for detecting glaucomatous visual field defects. This study evaluated the influence of sampling frequency on eye movement parameters in detecting glaucomatous visual field defects during a free-viewing task.

METHODS

We investigated eye movements in two sets of experiments: (a) young adults with and without simulated visual field defects and (b) glaucoma patients and age-matched controls. In Experiment 1, we recruited 30 healthy volunteers. Among these, 10 performed the task with a gaze-contingent superior arcuate (SARC) scotoma, 10 performed the task with a gaze-contingent biarcuate (BARC) scotoma and 10 performed the task without a simulated scotoma (NoSim). The experimental task involved participants freely exploring 100 images, each for 4 s. Eye movements were recorded using the LiveTrack Lightning eye-tracker (500 Hz). In Experiment 2, we recruited 20 glaucoma patients and 16 age-matched controls. All participants underwent similar experimental tasks as in Experiment 1, except only 37 images were shown for exploration. To analyse the effect of sampling frequency, data were downsampled to 250, 120 and 60 Hz. Eye movement parameters, such as the number of fixations, fixation duration, saccadic amplitude and bivariate contour ellipse area (BCEA), were computed across various sampling frequencies.

RESULTS

Two-way ANOVA revealed no significant effects of sampling frequency on fixation duration (simulation, p = 0.37; glaucoma patients, p = 0.95) and BCEA (simulation, p = 0.84; glaucoma patients: p = 0.91). BCEA showed good distinguishability in differentiating groups across different sampling frequencies, whereas fixation duration failed to distinguish between glaucoma patients and controls. Number of fixations and saccade amplitude showed variations with sampling frequency in both simulations and glaucoma patients.

CONCLUSION

In both simulations and glaucoma patients, BCEA consistently differentiated them from controls across various sampling frequencies.

Comparison of unifocal, flicker, and multifocal pupil perimetry methods in healthy adults

To this day, the most popular method of choice for testing visual field defects (VFDs) is subjective standard automated perimetry. However, a need has arisen for an objective, and less time-consuming method. Pupil perimetry (PP), which uses pupil responses to onsets of bright stimuli as indications of visual sensitivity, fulfills these requirements. It is currently unclear which PP method most accurately detects VFDs. Hence, the purpose of this study is to compare three PP methods for measuring pupil responsiveness.

Unifocal (UPP), flicker (FPP), and multifocal PP (MPP) were compared by monocularly testing the inner 60 degrees of vision at 44 wedge-shaped locations. The visual field (VF) sensitivity of 18 healthy adult participants (mean age and SD 23.7 ± 3.0 years) was assessed, each under three different artificially simulated scotomas for approximately 4.5 minutes each (i.e. stimulus was not or only partially present) conditions: quadrantanopia, a 20-, and 10-degree diameter scotoma.

Stimuli that were fully present on the screen evoked strongest, partially present stimuli evoked weaker, and absent stimuli evoked the weakest pupil responses in all methods. However, the pupil responses in FPP showed stronger discriminative power for present versus absent trials (median d-prime = 6.26 ± 2.49, area under the curve [AUC] = 1.0 ± 0) and MPP performed better for fully present versus partially present trials (median d-prime = 1.19 ± 0.62, AUC = 0.80 ± 0.11).

We conducted the first in-depth comparison of three PP methods. Gaze-contingent FPP had best discriminative power for large (absolute) scotomas, whereas MPP performed slightly better with small (relative) scotomas.

Developing a Screening Tool for Areas of Abnormal Central Vision Using Visual Stimuli With Natural Scene Statistics

Purpose

Previous studies show that some visual field (VF) defects are detectable from visual search behavior; for example, when watching video. Here, we developed and tested a VF testing approach that measures the number of fixations to find targets on a background with spatial frequency content similar to natural scenes.

Methods

Twenty-one older controls and 20 people with glaucoma participated. Participants searched for a Gabor (6 c/°) that appeared in one of 25 possible locations within a 15° (visual angle) 1/f noise background (RMS contrast: 0.20). Procedure performance was assessed by calculating sensitivity and specificity for different combinations of control performance limits (p = 95%, 98%, 99%), number of target locations with fixations outside control performance limits (k = 0 to 25) and number of repeated target presentations (n = 1 to 20).

Results

Controls made a median of two to three fixations (twenty-fifth to seventy-fifth percentile: two to four) to locate the target depending on location. A VF was flagged “abnormal” when the number of fixations was greater than the p = 99% for k = 3 or more locations with n = 2 repeated presentations, giving 85% sensitivity and 95.2% specificity. The median test time for controls was 85.71 (twenty-fifth to seventy-fifth percentile: 66.49–113.53) seconds.

Conclusion

Our prototype test demonstrated effective and efficient screening of abnormal areas in central vision.

Translational Relevance

Visual search behavior can be used to detect central vision loss and may produce results that relate well to performance in natural visual environments.

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

Purpose: There is a need for more intuitive perimetric screening methods, which can also be performed by elderly people and children currently unable to perform standard automated perimetry (SAP). Ideally, these methods should also be easier to administer, such that they may be used outside of a regular clinical environment. We evaluated the suitability of various methodological and analytical approaches for detecting and localizing VFD in glaucoma patients, based on eye movement recordings. Methods: The present study consisted of two experiments. In experiment 1, we collected data from 20 glaucoma patients and 20 age-matched controls, who monocularly viewed 28 1-min video clips while their eyes were being tracked. In experiment 2, we re-analyzed a published dataset, that contained data of 44 glaucoma patients and 32 age-matched controls who had binocularly viewed three longer-duration (3, 5, and 7 min) video clips. For both experiments, we first examined if the two groups differed in the basic properties of their fixations and saccades. In addition, we computed the viewing priority (VP) of each participant. Following a previously reported approach, for each participant, we mapped their fixation locations and used kernel Principal Component Analysis (kPCA) to distinguish patients from controls. Finally, we attempted to reconstruct the location of a patient's VFD by mapping the relative fixation frequency and the VP across their visual field. Results: We found direction dependent saccade amplitudes in glaucoma patients that often differed from those of the controls. Moreover, the kPCA indicated that the fixation maps of the two groups separated into two clusters based on the first two principal components. On average, glaucoma patients had a significantly lower VP than the controls, with this decrease depending on the specific video viewed. Conclusions: It is possible to detect the presence of VFD in glaucoma patients based on their gaze behavior made during video viewing. While this corroborates earlier conclusions, we show that it requires participants to view the videos monocularly. Nevertheless, we could not reconstruct the VFD with any of the evaluated methods, possibly due to compensatory eye movements made by the glaucoma patients.

Patients Prefer a Virtual Reality Approach Over a Similarly Performing Screen-Based Approach for Continuous Oculomotor-Based Screening of Glaucomatous and Neuro-Ophthalmological Visual Field Defects

Standard automated perimetry (SAP) is the gold standard for evaluating the presence of visual field defects (VFDs). Nevertheless, it has requirements such as prolonged attention, stable fixation, and a need for a motor response that limit application in various patient groups. Therefore, a novel approach using eye movements (EMs) - as a complementary technique to SAP - was developed and tested in clinical settings by our group. However, the original method uses a screen-based eye-tracker which still requires participants to keep their chin and head stable. Virtual reality (VR) has shown much promise in ophthalmic diagnostics - especially in terms of freedom of head movement and precise control over experimental settings, besides being portable. In this study, we set out to see if patients can be screened for VFDs based on their EM in a VR-based framework and if they are comparable to the screen-based eyetracker. Moreover, we wanted to know if this framework can provide an effective and enjoyable user experience (UX) compared to our previous approach and the conventional SAP. Therefore, we first modified our method and implemented it on a VR head-mounted device with built-in eye tracking. Subsequently, 15 controls naïve to SAP, 15 patients with a neuro-ophthalmological disorder, and 15 glaucoma patients performed three tasks in a counterbalanced manner: (1) a visual tracking task on the VR headset while their EM was recorded, (2) the preceding tracking task but on a conventional screen-based eye tracker, and (3) SAP. We then quantified the spatio-temporal properties (STP) of the EM of each group using a cross-correlogram analysis. Finally, we evaluated the human-computer interaction (HCI) aspects of the participants in the three methods using a user-experience questionnaire. We find that: (1) the VR framework can distinguish the participants according to their oculomotor characteristics; (2) the STP of the VR framework are similar to those from the screen-based eye tracker; and (3) participants from all the groups found the VR-screening test to be the most attractive. Thus, we conclude that the EM-based approach implemented in VR can be a user-friendly and portable companion to complement existing perimetric techniques in ophthalmic clinics.