Deviated Saccadic Trajectory as a Biometric Signature of Glaucoma

Abstract

Purpose: To investigate whether the trajectories of saccadic eye movements (SEMs) significantly differ between glaucoma patients and controls.

Methods: SEMs were recorded by video-based infrared oculography in 53 patients with glaucoma and 41 age-matched controls. Participants were asked to bilaterally view 24°-horizontal, 14°-vertical, and 20°-diagonal eccentric Goldmann III-sized stimuli. SEMs were evaluated with respect to the saccadic reaction time (SRT), the mean velocity, amplitude, and two novel measures: departure angle (DA) and arrival angle (AA). These parameters were compared between the groups and the associations of SEM parameters with glaucoma parameters and integrated visual field defects were investigated.

Results: Glaucoma patients exhibited increased mean SRT, DA, and AA values compared with controls for 14°-vertical visual targets (P = 0.05, P < 0.01, and P < 0.01, respectively). The SRT, DA, and AA were significantly associated with the mean and pattern standard deviations of perimetry and with the mean RNFL thickness by OCT (all P < 0.001). Glaucoma was associated with the AA (P = 0.05) and both the SRT (P = 0.01) and DA (P = 0.04) were associated with integrated visual field defects.

Conclusions: The saccadic trajectories of glaucoma patients depart in an erroneous path and compensate the disparity by deviating the trajectory at arrival.

Translational Relevance: The initial deviation that we observed (despite continuous exposure to the stimulus) suggests the disoriented spatial perception of glaucoma patients which may be relevant to difficulties encountered daily.

Introduction

Glaucoma is a multifactorial, progressive optic neuropathy that leads to gradual and irreversible visual field loss. Glaucoma is a leading cause of acquired blindness worldwide and is characterized by axonal degeneration that affects the afferent visual pathway (i.e., from retinal ganglion cells to the lateral geniculate nucleus and the visual cortex).^1–3 Peripheral vision is most susceptible to glaucomatous damage, substantial changes are evident in periphery before any loss of central visual acuity.⁴ Nonetheless, patients with glaucomatous optic neuropathy (GON) experience difficulties with everyday activities and have an increased risk of falls.^5–9 

The human visual acuity is highest for images that reach the fovea where the photoreceptor density is greatest and lowest for images that reach peripheral retinal regions.^10,11 To maintain adequate vision of the surrounding environment, sufficient perception of peripheral targets and a swift gaze shift are needed to bring the fovea to a position where it can evaluate the area of interest.¹² The saccadic system calculates the distance and direction of a peripheral target from the current gaze position, then generates high-velocity movements of both eyes that bring the image onto or near the fovea.¹³ Therefore, the struggles of daily living may be affected not only by poor perception of peripheral targets, but also because of a miscarried saccadic system that fails to properly shift gaze. 

The saccadic oculomotor system has been extensively studied because saccadic eye movements (SEMs) can be accurately measured,^14–16 the neurons that control saccades are readily accessible to microelectrodes,^17–19 and the neural network underlying saccade generation is well-known.^12,13 Reliable SEM parameters include the saccadic latency or saccade reaction time (SRT),²⁰ saccadic duration,²¹ amplitude,²² and peak velocity.²³ The SRT is increased in patients with various optic nerve pathologies that affect the retinal ganglion cells necessary to convey visual signals to the saccade-generating network.^24–26 These parameters are also used to assess the integrity of the saccade-generating neural network in patients with various brain diseases.^27–29 

Most recent eye movement studies have used high-speed video-based oculography. Because up to 1000 eye movement coordinates may be registered per second, saccades are often simplified to straight lines that connect the start and end points of an eye movement above a certain velocity and acceleration.³⁰ However, the real trajectory may not always be a straight line, and some of the movement path can be curved or irregular. In particular, it is known that saccades curve away from covertly attended locations and even from visual distractors, and the magnitude of such curvature is associated with the saliency of the distractor.^31,32 Based on this speculation, we investigated whether SEMs deviated from the intended paths in subjects with GON. We focused on whether the trajectory changed after commencement of a saccade, which might suggest distorted spatial perception of an intended target. Therefore, the saccade departure angle (DA) and arrival angle (AA) that measures deviation of the path from the intended target were quantified. The angles between the intended target and the actual trajectory throughout an SEM set were measured in 10 different temporal sections. Conventional saccade parameters (e.g., SRT, mean and peak velocities, and amplitude) were also acquired and compared among normal controls and glaucoma patients. 

Materials and Methods

After the study protocol had been approved by the Catholic University Incheon St. Mary's Hospital Research Ethics Board, informed consent was obtained from glaucoma patients and age-matched controls between the ages of 26 and 84 years. Patients with GON (n = 53) were prospectively recruited from the practice of a glaucoma specialist (Y.C.K.) from October 2021 to May 2022. GON was defined by characteristic changes in the optic nerve head and thinning of the retinal nerve fiber layer (RNFL), with corresponding visual field changes determined by white-on-white static automated perimetry (Carl Zeiss Meditec, Jena, Germany). The exclusion criteria were all other causes of secondary glaucoma; any nonglaucomatous eye disease; incisional eye surgery within the previous 1 month; central visual acuity worse than 0.2 logarithm of the minimum angle of resolution; any history of neurological disease, psychiatric condition, or neurological disorder (including cognitive impairment or dementia); and any use of psychotropic medications known to affect saccade velocity.³³ Age-matched controls (n = 41) were recruited from among patients who were undergoing follow-up because of high cup-to-disc ratios but otherwise exhibited normal findings during consecutive eye examinations. GON patients and age-matched controls underwent comprehensive clinical eye examinations including Landolt C chart–assisted measurements of logarithm of the minimum angle of resolution best-corrected visual acuity, slit-lamp biomicroscopy, axial length measurement via ocular biometry (IOL Master; Carl Zeiss Meditec), digital color fundus photography (VX-10i camera, Kowa Co., Nagoya, Japan), and optical coherence tomography (DRIOCT Triton; Topcon Corporation, Tokyo, Japan). The study conformed to the Declaration of Helsinki, and all data were anonymized before transfer to a secure computer database at the university. 

Eye Movement Recordings

All glaucoma patients and controls were tested under constant luminance of 300 cd/m² (LS-150 Luminance Meter, Osaka, Japan). Each participant was seated on a chair with the head stabilized by a chinrest at 60 cm from a light-emitting diode backlit display monitor (59.8 × 33.6 cm; 27 inches; 16:9 aspect ratio) with a resolution of 1920 × 1080 pixels and a refresh rate of 75 Hz (LG 27MK430H, LG Electronics, Seoul, Korea). Binocular eye movements were simultaneously recorded using an EyeLink 1000 Plus device (SR Research Ltd., Mississauga, Ontario, Canada) that recorded eye position coordinates at 1000 Hz. The mean visual angle accuracy exceeded 0.5° over the trackable range of 32° × 25°. The default saccade detection thresholds were velocity greater than 30°/s and acceleration greater than 4000°/s². Before testing, calibration (13 positions) was performed; the standard was required to be good, as defined by the manufacturer. A drift check was performed between each trial; if drift was substantial, recalibration was conducted automatically. 

Trial Sequence and Saccade Stimulus

Trial sequence is described in Figure 1. (i) Participants were instructed to fixate on a 0.43°-diameter, circular fixation point (equivalent to Goldmann size III, the standard stimulus of the Humphrey Visual Field Analyzer) at the center of the monitor. (ii) After random period of 1500 to 3000 ms, the fixation point disappears and simultaneously a peripheral target appears. Subjects are instructed to perform a single saccade to the target as soon as possible. (iii) After subjects fixated on the target for 5000 ms, the target point disappears and simultaneously the central fixation point reappears at the same location. (iv) The fixation point is presented for random period of 1500 to 3000 ms and another peripheral target in the random order appears with the fixation point disappearing simultaneously. (v) The trials are repeated 16 times at each of 8 locations. 

Figure 1.

 

Schematic diagram showing the trial sequence. (i) A 0.43°-diameter, circular fixation point appears in the center of the screen and subjects were instructed to fixates on the point. (ii) After random period of 1500 to 3000 ms, the fixation point disappears and simultaneously a peripheral target appears. Subjects are instructed to perform a single saccade to the target as soon as possible. (iii) After subjects fixated on the target for 5000 ms, the target point disappears and simultaneously central fixation point reappears at the same location. (iv) The fixation point is presented for random period of 1500 to 3000 ms and another peripheral target in the random order appears with the fixation point disappearing simultaneously. (v) The trials are repeated 16 times at each of 8 locations.

The stimulus had a luminance of 127 cd/m² (equivalent to 14 dB on the Humphrey Visual Field Analyzer luminance scale) against a background luminance of 10 cd/m². Figure 2 depicts the location of the saccade stimuli (blue dots) superimposed on the placement of the standard visual field test (black). Eight stimulus locations at which glaucomatous arcuate scotoma frequently occur were designated as stimulus points. The stimulus was a 0.43°-diameter located at 24° of horizontal eccentricity, 14° of vertical eccentricity, and 20° of diagonal eccentricity which corresponded to the placement of the standard automated perimetry (Figure 2). Participants were instructed to look toward the presenting stimulus as rapidly and accurately as possible after presentation. 

Figure 2.

 

Schematic diagram showing the location of the saccade stimuli (blue dots) superimposed on the placement of the standard visual field test (black). Each black numbered square represents a visual field test stimulus point in the respective left and right eye format. Eight sites at which glaucomatous arcuate scotoma frequently occur were designated as stimulus points. A 0.43°-diameter stimulus at 24° of horizontal eccentricity, 14° of vertical eccentricity, and 20° of diagonal eccentricity appeared for 5000 ms.

Integrated Binocular Visual Fields

Integrated visual field (IVF) was calculated by choosing the higher sensitivity at two overlapping locations on each monocular Humphrey visual field (HVF) test using the Swedish interactive thresholding algorithm 24-2 (Humphrey Visual Field Analyzer; Carl Zeiss Meditec, Dublin, CA, USA) as described previously.³⁴ The IVF determined whether the participant could identify the stimulus points. A point was labeled the IVF defect (IVFD) if the P value of both pattern standard deviation (PSD) was less than 1%. Each IVF point corresponding with a point of the saccade stimulus was identified (Supplementary Table 1). A reliable visual field was defined as a field with less than 20% fixation loss, as well as false-negative and false-positive rates of less than 15%. 

Data Analysis

Saccadic trajectory analysis was performed using a customized algorithm (iDynamics, Seoul, Korea) that automatically detected the start and end points of all saccades from both eyes, then measured the saccadic trajectories between the onset and the offset. To better define the onset and offset of the saccadic trajectory, it is necessary to extract the hesitating eye movements at the start of the saccade and the wandering eye movements after the saccade ends. Therefore, the algorithm considered the starting direction (within 90°) of the saccade along with the minimum size (>6° amplitude) and speed (>30°/s velocity). The offset of the saccade was defined as the most advanced time within 0.2 seconds after finding the peak velocity with the largest magnitude and velocity. 

Any trial that lacked data concerning more than one-third of a saccadic progression was excluded. Otherwise, missing data were interpolated. The data from each trial (eight different positions) were converted to single x and y coordinates by affine transform generalization (Equation 1). The x axis was a straight path that connected the start point to the intended target, and the y axis was perpendicular to the x axis. All SEMs were converted into points on these axes, regardless of stimulus position (Equation 2).  

\begin{eqnarray}\left[ {\frac{{x^{\prime}}}{{y^{\prime}}}} \right] = {\rm{\;}}\left[ {\begin{array}{@{}*{2}{c}@{}} {cos\theta }&{sin\theta }\\ { - sin\theta }&{cos\theta } \end{array}} \right]\left[ {\frac{x}{y}} \right] - {\rm{\;}}\left[ {\frac{{{x_{target\left( {n - 1} \right)}}}}{{{y_{target\left( {n - 1} \right)}}}}} \right]\end{eqnarray}

(1)

 

\begin{eqnarray}\theta = ta{n^{ - 1}}\left( {\frac{{{y_{target\left( n \right)}} - {y_{target\left( {n - 1} \right)}}}}{{{x_{target\left( n \right)}} - {x_{target\left( {n - 1} \right)}}}}} \right)\end{eqnarray}

(2)

 

An SEM was divided into 10 equidistant intervals, each of which included k scanpaths; each scanpath position was recorded over 0.001 second. The maximum tangent angles formed by all k scanpaths from the starting points within all sections were calculated (Equation 3).  

\begin{eqnarray}&& {x_n} = {x_{min}} + n\left( {\frac{{{x_{max}} - {x_{min}}}}{{10}}} \right) \left[ {n = 0{\rm{\;}}\ \sim {\rm{\;}}10} \right] \nonumber \\ && Tangent{\rm{\;}}Angl{e_{n + 1}} \nonumber \\ && = max\left( {{\rm{\;}}\left| {ta{n^{ - 1}}\left( {\frac{{{y_k} - {y_{{k_0}}}}}{{{x_k} - {x_{{k_0}}}}}} \right)} \right|} \right)\left[ {{x_n}{\rm{\;}}\sim {\rm{\;}}{x_{n + 1}}} \right] \end{eqnarray}

(3)

 

The maximum tangent angle of the first interval was designated the DA; the maximum tangent angle of the last interval was designated the AA (Figure 3). The SRT, mean and peak velocities, amplitude, and gain (i.e., the ratio of the amplitude of the saccade to the amplitude of the peripheral visual target) of each saccade were acquired using the default settings of EyeLink DataViewer (SR Research Ltd.). The accuracy of automated saccade detection was verified by experienced investigators (Y.J.S. and Y.C.K.). 

Figure 3.

 

Measurement of saccadic deviation angle. A saccade was divided into 10 equidistant intervals and the maximum tangent angles formed from the starting points within all sections were calculated. The maximum tangent angle of the first interval was designated the DA and the maximum tangent angle of the last interval was designated the AA.

The demographic and comprehensive ophthalmic data of all participants who underwent saccade analyses were compared using the Mann–Whitney U test. Associations between saccadic and glaucoma parameters were calculated using the point-biserial method. We used univariate and multivariate logistic regression analyses to identify factors associated with glaucoma; we calculated adjusted odds ratios with 95% confidence intervals. Five independent variables were tested for the correlation and logistic regression analyses, the adjusted P value to reject the null hypothesis was adjusted to 0.01. A P value of less than 0.05 was considered statistically significant in the Mann–Whitney U test comparison. Demographic and ophthalmic data are presented as medians with interquartile ranges. Statistical analyses were performed and figures were prepared using Python software. 

Results

We included 53 GON patients and 41 age-matched healthy controls with mean ages of 54.60 ± 14.11 years and 56.02 ± 12.36 years, respectively (P = 0.28). The sex ratio, best-corrected visual acuity, and axial length did not significantly differ between the groups. However, differences were apparent in terms of the HVF mean deviation (HVF-MD), HVF-PSD, and the global, superior, and inferior RNFL thicknesses (all P < 0.001) (Table 1). We compared 5746 saccades of the GON group with 3990 saccades of the control group by saccadic direction and eccentricity (Table 2 and Supplementary Figure 1). Although there were various differences according to eccentricity and direction, the difference between the two groups was particularly large in the DA and AA at the 14 degree vertical saccade (both P < 0.001). There was also difference in amplitude and SRT by different eccentricity and direction. 

Table 1.

 

Demographics and Ocular Clinical Characteristics of Patients With Glaucoma and Healthy Control Groups

 

Table 2.

 

Comparison of Saccadic Characteristics in Glaucoma and Healthy Control Groups by Different Eccentricity and Direction

 

We tested whether the saccade parameters were associated with the conventional glaucoma parameters of the respective eyes (Table 3). Surprisingly, the HVF-MD and HVF-PSD were significantly associated with saccadic trajectory parameters, including the DA and AA (all P < 0.01). The global RNFL thickness was significantly associated with the DA and AA (P < 0.01 and P = 0.01, respectively). The SRT was associated with the HVF-MD, HVF-PSD, and global RNFL thickness (all P < 0.01). However, the amplitude and average velocity were not significantly associated with most of glaucoma parameters. The extents of association differed according to saccade direction; the saccadic trajectory parameters exhibited stronger associations in the upward and downward directions, whereas the horizontal and diagonal saccades were minimally associated with glaucoma parameters (Supplementary Table 2). Univariate analysis revealed that the DA and AA were significantly associated with glaucoma status (P = 0.020 and P = 0.005, respectively); the AA association remained significant on multivariate analysis (P = 0.048) (Table 4). Table 5 explores whether each saccade involved IVFDs at the corresponding stimulus point. Because tests proceeded simultaneously for both eyes, the presence of an IVFD implied that both eyes exhibited a VFD at the same location. We sought associations between the presence of an IVFD and the trajectory parameters. In contrast with the trajectory parameters associated with glaucoma, only the DA was significantly associated with the presence of an IVFD (P = 0.02) (Table 5). On univariate analysis according to IVFD status, the DA and SRT were significantly associated with defects (P = 0.023 and P = 0.003, respectively). On multivariate analysis, the increases in the SRT and DA remained statistically significant (P = 0.006 and P = 0.043, respectively). 

Table 3.

 

Associations Between the SEMs Parameters With the Glaucoma Parameters

 

Table 4.

 

Saccadic Characteristics Associated With Glaucoma.

 

Table 5.

 

Saccadic Characteristics Associated With IVFDs

 

Figure 4A shows the MDs between the intended paths and actual eye movements over the 10 sections. Throughout the trajectory, glaucomatous SEMs exhibited significantly larger deviations from the intended paths in nine of the 10 sections. Figure 4B compares the MDs of glaucoma patients without and with IVFDs. Although the differences in the middle of the saccadic trajectories were not statistically significant, the DA and AA significantly differed (P = 0.02 and P = 0.02, respectively) (Supplementary Table 3). Notably, the DA and AA deviations were opposite in nature; patients without IVFDs exhibited a significantly larger AA compared with patients with IVFDs, whereas patients with IVFDs exhibited a significantly larger DA compared with patients without IVFDs. The glaucoma patients and controls were compared in terms of saccadic directions; no significant differences in amplitude, mean or peak velocity, duration, or gain were apparent. However, in terms of the upward saccades, significant differences in the DA and AA parameters were evident (P = 0.01 and P = 0.05, respectively). These trajectory differences were not statistically significant in either horizontal direction. The diagonal saccades exhibited intermediate results in terms of the trajectory parameters; the AAs of the right-downward saccades significantly differed (P = 0.01) (Figure 5). 

Figure 4.

 

Mean deviations between the intended paths and actual eye movements over the 10 sections. (A) Throughout the trajectory, glaucomatous SEMs exhibited significantly larger deviations from the intended paths in nine of the 10 sections. (B) Notably, the DA and AA deviations were opposite in nature; patients without IVFDs exhibited a significantly larger AA compared with patients with IVFDs, whereas patients with IVFDs exhibited a significantly larger DA compared with patients without IVFDs.

Figure 5.

 

Comparison of saccadic parameters with respective directions. No significant differences in amplitude, mean or peak velocity, duration, or gain were apparent between control (A) and glaucoma (B) group. However, significant differences in the DA and AA parameters were evident in upward and downward saccades.

Discussion

Our results show that, in a randomized sequence of saccades, SEMs from a GON patient departs in an erroneous path and compensates the disparity by also deviating the trajectory at its arrival compared with age-matched controls. The magnitudes of the deviations were associated with certain glaucoma parameters in terms of HVF and optical coherence tomography. Moreover, our trajectory analysis shows that the glaucomatous saccades have their trajectory significantly deviated from the beginning of its initiation. Our results imply that glaucomatous SEMs without IVFDs actively correct their final trajectories to the intended targets, whereas saccades with IVFDs at the corresponding points exhibit less ability to correct these trajectories. Taken together, our findings suggest that a deviated saccadic trajectory is an inherent characteristic of glaucoma; they may explain the difficulties encountered by glaucoma patients during daily life. 

It has been suggested that SEMs are generated in the cortical eye fields (primary visual, parietal, frontal, and supplementary) and the subcortical network structures (superior colliculus, thalamus, and striatum).³⁵ Therefore, it can be assumed that malfunctions in these neural networks may cause dysfunctional saccades. Among others, the superior colliculus is presumably the origin of such malfunctions. The superior colliculus is the principal conduit of the output saccadic stream to the oculomotor complex; lesions in the superior colliculus eliminate the stream.^36,37 Aizawa et al. found that the injection of muscimol (a GABA receptor agonist) changed saccadic trajectories, such that they became consistently curved and slower, with longer latencies, which is equivalent to our findings regarding glaucomatous SEMs.^18,36,38 The anatomical details of direct optic nerve projections to the dorsal midbrain and the superior colliculus have been described as well.¹⁹ Thus, a glaucomatous optic nerve may affect SEM generation, such that the saccadic trajectory becomes miscarried. 

To our knowledge, we are the first to suggest that changes in saccadic trajectory constitute an intrinsic feature of glaucomatous SEMs. Previous studies described reduced saccade velocity,³⁹ hypometric saccade amplitude,²³ and delayed SRTs.^20,25 However, our data clearly imply that an initial saccade deviation is a hallmark of glaucomatous saccades. This hypothesis is supported by proportional associations between the extent of the deviated angle and the glaucoma parameters in terms of HVF and optical coherence tomography. This hypothesis is also consistent with the spatiotemporal properties of glaucomatous SEMs. Soans et al.⁴⁰ reported that the increased spatiotemporal properties of vertical SEMs clearly distinguished glaucoma patients from healthy individuals. Their spatiotemporal parameters considered the spatial errors and temporal delays between a stimulus and the SEM trajectory, analogous to the angular deviation measured in our study. Notably, their study showed more prominent upward and downward saccades, consistent with our observations. Furthermore, studies that showed reduced accuracies of glaucomatous SEMs (compared with normal SEMs) may be consistent with our hypothesis; a dysfunctional trajectory may end at an erroneous fixation point.⁴¹ Therefore, the deviated trajectories of glaucomatous SEMs that we report may not reflect a specific experimental situation; they may be inherent features of glaucomatous SEMs. 

Our observations have implications in terms of spatial perception by GON patients. Thus far, functionality in glaucoma patients has usually been evaluated using the HVF. This approach only explores whether a patient recognizes a static stimulus when the eye is stationary. The HVF test does not evaluate how a patient perceives the location of a stimulus. Our data suggest that glaucoma patients not only experience VFDs, but also have disoriented spatial perception. The presence of an angular deviation even at the commencement of saccade generation suggests that the patient is aware of a stimulus, but its perceived location may be inaccurate. This hypothesis is supported by the eye movement perimetry measurements collected when SEM directional biases were imposed during the exploration of stimulus recognition.⁴² We presented stimuli for 5 seconds, whereas the eye movement perimetry stimulus exposure time in the cited work was 0.2 second. The SRT of a typical SEM is 0.2 second; this is the time before the stimulus disappears. Thus, the SEMs of the eye movement perimetry study proceeded to positions at which the patient remembered that a stimulus had been located, creating a directional bias that revealed perception of the spatial orientation. The initial SEM deviation that we observed (despite continuous exposure to the stimulus) emphasizes the disoriented spatial perception. This theoretical framework is consistent with the proposed paradigm of a decreased spatial cognitive map during explorations of virtual reality,⁴³ which may provide additional explanation for the decreased mobility and poor general performance in glaucoma patients.^44,45 

According to our paradigm, the deviated trajectory is an intrinsic glaucomatous feature based on disoriented spatial perception. However, it could be hypothesized that the deviation is caused by a wandering gaze, failing to locate the target because of a VFD at the corresponding point. Our comparison of the saccade parameters of glaucoma patients with and without IVFDs suggests that, although the DAs are similar, the AAs significantly differ. The AA increased in patients without IVFDs, whereas the DA increased in patients with IVFDs. These findings suggest that an erroneous departure was actively compensated in patients without IVFDs; patients with IVFDs may have limited capacity to compensate for such a departure. Moreover, an increased SRT was associated with SEMs that involved IVFDs, but not with SEMs that lacked IVFDs, perhaps indicating that SEMs that lacked IVFDs were able to perceive the target, which also indicate the neural capacity to restore the trajectory to an appropriate path. 

Daga et al.⁴³ recently examined the wayfinding abilities of glaucoma patients in a virtual reality environment. The patients found it more difficult to recognize targets in a virtual room with multiple visual cues compared with a room with fewer such cues. The findings suggested that glaucoma patients’ visual searching was ineffective, which disrupted patients’ ability to build a detailed spatial cognitive map. The authors suggested eye tracking methods when exploring deficient behaviors with respect to wayfinding. These observations are consistent with our hypothesis: glaucoma patients exhibit poor visual field sensitivity and have deviated perception of spatial orientation as well. Extensive visual information must be integrated when recognizing a three-dimensional space; the cognitive map may become distorted by the accumulation of deviated perceptions. 

Our work had some limitations. First, although the simulated IVF values were in good agreement with the values generated by the binocular Esterman visual field test,³⁴ these values may differ when a bilateral gaze is involved, thereby leading to incorrect VFDs. Additionally, head stabilization by a chinrest also differs from realistic visual conditions, because compensations such as head movements are not allowed. We plan to remove this restriction in future studies. Second, we could not exclude the possible effects of glaucoma medications on SEMs. However, to our knowledge, there have been no suggestions that prostaglandin analogs, alpha agonists, or beta-blockers affect SEMs. Further investigations with additional patients are needed. Third, a cathode ray tube monitor is often used during eye tracking experiments; such monitors exhibit fast reaction and response times, and they permit large viewing angles. A liquid crystal display is slower and the viewing angle is smaller. A light-emitting diode monitor (such as the monitor used in the present study) exhibits a large viewing angle (≤178°) and an adequate response time. Fourth, all stimuli were two-dimensional and not in full color; thus, they did not reflect the dynamic nature of the real world. Additionally, the patients rarely move their eyes by more than 15° in natural conditions. Because we maintained constant saccadic amplitudes in all directions, the saccadic trajectories at various amplitudes should be evaluated in future studies. 

In conclusion, when binocular saccades were induced, patients with glaucoma exhibited consistently different eye movements, compared with healthy controls. Specifically, the saccadic trajectories of glaucoma patients departed erroneously and the disparities were compensated by deviating at its arrival. These between-group differences were associated with common clinical measures of glaucoma; the DA and AA were associated with changes in HVF-MD and in optical coherence tomography–based measurements of RNFL thickness. The initial changes may not indicate wandering; SEMs in patients without VFDs were actively corrected in terms of the final trajectories, but the SEMs of patients with VFDs exhibited less correction. We present a novel analysis of saccadic trajectory. Our results may be used to analyze the eye movements of glaucoma patients and elucidate the challenges that they experience each day. 

Acknowledgments

Supported by National Research Foundation of Korea (NRF) grant funded by the Ministry of Education (RS-2022-00167024) and the Translational R&D Project through Institute for Bio-Medical convergence, Incheon St. Mary's Hospital, The Catholic University of Korea (IBC-2022M-09). The funding organizations had no role in the design or conduct of this research. 

YCK is listed as inventors on the Korean patent application “Apparatus and Method for Determining Glaucoma,” which is partially based on the method described in this manuscript. 

Disclosure: J.S. Yeon, None; H.N. Jung, None; J.Y. Kim, None; K.I. Jung, None; H.-Y.L. Park, None; C.K. Park, None; H.W. Kim, None; M.S. Kim, None; Y.C. Kim, None