Saccades are fast eye movements to a target position that bring objects of interest to the central visual focus (1, 2) and orient the gaze towards an object of interest. These may be horizontal, vertical, or oblique. However, most saccades are horizontal or near-horizontal as reported by van der Linde et al. (3). They can be voluntary, when scanning a page of text, reflexive to a new visual stimulus, or involuntary during the fast phase of nystagmus.

Parkinson’s disease (PD) is a neurodegenerative disease that affects about 1% of the population above the age of 60 (4). Several oculomotor alterations have been associated with PD. Saccadic hypometria has been found in PD as shown by Pretegiani et al. (5). Many researchers have observed eye saccades features as a potential indication to both diagnose PD and evaluate its progression and treatment effects (1, 6–11).

Basic saccadic circuit involves the cortical and subcortical areas of the brain. The superior colliculus (SC), located in the midbrain, plays a central role in saccadic movements and eye fixation (1, 2, 12–16). The SC receives inputs from the retina, as well as the basal ganglia (BG) and various cerebral cortical regions. This initiates saccadic movements with bursts of neuronal activity that are transmitted to oculomotor control areas. Upon reaching the target, neural integrators generate a tonic burst of activity to keep the eye in a position (2, 17, 18). The burst neurons are then inhibited by omnipause neurons to end the saccadic movement. Although their saccadic neural circuits are similar, the oculomotor control pathways for horizontal and vertical saccades are separate; paramedian pontine reticular formation (PPRF) is recruited for horizontal movements, and the midbrain tegmentum for vertical (2, 19–21). Saccadic initiation is triggered by visual inputs from the optic nerves and modified by motor planning and cognitive inputs from the basal ganglia (BG) and frontal cortical areas. The frontal cortex and cerebellum also have direct influences on pontine burst cells. The dopaminergic deficiency associated with PD would be expected to affect BG control of the saccadic neural circuitry.

Saccadic function can be tested at several levels (1, 9). Reflexive saccades primarily assess motor function, where the participants follow their gaze to a visual target. The other option is an anti-saccadic task (9), which is a voluntary saccade. Instructions are given to move in the opposite direction from the visual target, thus evaluating motor and cognitive functions together with the capacity for inhibitory control. Another is the memory-guided saccadic tasks, which assess spatial working memory, and require rapid eye movements to previously remembered target locations (22). Predictive saccadic tasks, where the subject is expected to move their eyes to a predicted target, assess higher cognitive functions.

Many researchers have reported abnormality in PwPD saccade parameters. Lueck et al. (23) observed abnormality in PwPD saccade movements during memory-guided saccadic tasks, while no impairments were observed in anti-saccadic tasks. Briand et al. (24) did not find any impairment in the reflexive task but PwPD were found to have more latency and made a large percentage of errors compared to the healthy subjects in a voluntary saccade task. Blekher et al. (25) investigated moderate to advanced PD patients and found that voluntary saccades were more impaired in advanced PD as compared to moderate PD. Chan et al. (26) found that PwPD made more directional errors in anti-saccade and memory-guided tasks in comparison to healthy subjects. However, a study conducted by Amador (27) did not show significant differences between PwPD and AMC in anti-saccade tasks. Instead, they reported evidence of execution deficits in PwPD during voluntary and inhibiting reflexive saccades. Van Stockum (28) observed reduced saccade latencies in PwPD during reflexive tasks with the discriminator. Briand (6) detected abnormalities in visually guided and voluntary saccadic movement of PwPD. Waldthaler (29), who observed the parameters of horizontal and vertical reflexive saccades and anti-saccades in 40 patients found that medication did not alter saccade amplitudes and had opposing effects on the initiation of the visually guided saccade (reflexive saccade) and anti-saccade. Contrary to the other researchers, they found that medication such as levodopa increased the reflexive saccade latency.

From the above-mentioned studies, it is observed that PwPD have consistently been found to be more impaired when performing saccadic tasks that involved cognitive functions such as anti-saccades, memory-guided saccades and predictive saccades. However, there is inconsistency in the results for reflexive saccadic tasks, which assess the motoric exhibitory and inhibitory mechanism of BG (1). There may be many reasons under-pinning this such as differences in experimental protocols, equipment used, and participant factors. Thus further experimental investigation is neccessary to better understand the difference between AMC and PwPD on motoric function observed in reflexive saccadic tasks.

This paper reports an experimental investigation of the association of the reflexive eye saccadic parameters with PwPD. The eye gaze data was recorded for PwPD and AMC, and the test was repeated 20 times to determine the repeatability. The parameters were statistically analyzed to identify those with statistically significant differences between the two groups.

There are four major findings in this study related to the difference in the saccadic parameters between PwPD and AMC: (1) The saccadic response time of PwPD was faster than that of AMC in the case of reflexive saccadic movements when the target was traveling away from the center. (2) PwPD had higher inaccuracy in reaching the target. (3) PwPD had greater fluctuations with a steady target, and (4) the saccadic features were effective in differentiating PwPD and AMC.

Our study supports the findings of Chan et al. (26), who found that the latency among PwPD was less than those of AMC. They suggested that this was related to the deficit in the PwPD’s ability to inhibit automatic saccades (reflexive saccades initiated by the sudden appearance of a visual target). Briand et al. (6), after comparing the cued vs. uncued response time of PwPD vs. AMC, found that PwPD was faster when cued, but slower when uncued. They suggested that automatic spatial attention processes were more active in PwPD. The performance of PwPD on tasks involving reflexive spatial attention indicated the hyper-reflexivity in PwPD.

However, this is not universally accepted. There are inconsistencies in the results reported in the studies of PD patients’ eye movement on reflexive tasks (1, 11, 29). Some of the studies reported prolonged latency in PwPD when performing visually guided saccadic movements (5, 11, 34–37). This latency was linked to the loss of dopamine neurons resulting in the excessive inhibition of SC neurons by the BG, specifically the substantia nigra pars reticulata (SNr) (37, 38). By detecting and removing the invalid saccades has shown significant differences between AMC and PwPD, which may explain the inconsistencies in previous studies. A systematic literature review needs to be conducted, which will be useful in confirming the source of the inconsistencies.

One shortcoming in earlier research in this field was the small sample size. Our work with 70 participants (42 with PwPD and 28 AMC), is a significantly larger study than most of the similar investigations. We also removed those samples with invalid saccades and thus reduced the variability in the results. Our results support the findings of Chan (26), Brian (6), and Fooken (8), all of whom observed that PwPD participants exhibited more rapid express saccades in the pro-saccade task. Our PwPD were mostly in their early stage of the disease (PD duration of 5 years or less) (30) and this supports the argument made by Pretegiani (5) that reflexive saccades are typically preserved in the initial stages of PD. The latency of reflexive saccades can be maintained during the early stages of the disease, enabling PwPD patients to produce saccades even more rapidly than AMC individuals, particularly for small target eccentricities (5, 37).

The shorter latency in PwPD has also been mediated by higher-order cognitive processes such as attention (39–41). The LATER model (42–45) of saccadic eye movements relates saccadic latencies to the neural threshold to make the decision to move the eye, which is influenced by the level of attention. In our study, we observed that both groups had about 20% to 35% invalid saccades. While this result is consistent with the previous findings (5, 11, 37), however, in this study, the invalid saccade recordings were removed from the analysis to improve the quality of the data. Hence, we are unable to comment on this factor.

The accuracy of the saccadic movement of an individual is the ability to accurately reach the coordinate of the target. Some studies reported hypometric saccades of PD patients (5, 8, 26, 28, 46–48) when eye gaze did not reach or overshot the target position. The phenomenon may be related to excessive superior colliculus inhibition (37). Hypermetric cases in PwPD have been reported by Ba et al. (49), who linked this to the impairments in the inhibition of the reflexive saccade mechanism (37). Our study found that PwPD had a higher saccadic deviation (Sdev) compared to AMC.

Our results showed that there were significantly higher horizontal and vertical fluctuations in PwPD. This could explain why PwPD appears to lack the ability to focus on and scan a scene. It suggests that the tonic burst from the neural integrator (2), which is responsible for maintaining the eye in a fixed position, is impaired in PwPD.

This paper has shown that there are significant differences between the saccades of healthy people and those with Parkinson’s disease. However, it suffers from number of limitations which need to be addressed before its suitability can be confirmed. We have identified four major limitations, and list these below.

The first limitation is that while the selection of the 60Hz sampling was to make this suitable for easy access to clinicians, this has limited the choice of signal features and with a higher potential of error. The second limitation in this study is a gender imbalance in the data; the number of female participants for both, the PwPD and AMC groups is higher than male. While this was despite gender-unbiased participant recruitment, and the study by Wilson et al. (31) suggested there was no significant difference between males and females on any saccade parameters up to 35 degrees, nevertheless this needs to be addressed in future studies. The third shortcoming of this study is that it has only studied PwPD and age matched healthy control, but has not considered people with other neurological or ophthalmological disorders such as Atypical Parkinsonian or optic neuropathy such as glaucoma. These diseases could also alter the saccades which would make the results less specific. Another limitation is that while most of the PwPD were clinically identified to be in the early stage of the disease, the complete UPDRS-III score for some of them was not available. Further, the medicine details were also not available. Thus, the correlation between the severity of the disease, or influence of medication to the saccade parameters could not be investigated. Another limitation is that this is an observational case-control study; a longitudinal study to identify the changes to these parameters with the progression of the disease would be useful. Further, this study has only investigated the reflexive saccade. For a more complete understanding of the problem, voluntary tasks such as the antisaccadic, memory saccadic, and predictive saccades also need to be studied.

This work has the potential for being used to assist the neurologists perform diagnosis on people for detecting diagnosis. However, this needs to be tested for generalisability and for its suitability in the clinical work-flow to ensure it meets the clinical needs.

This study has determined that PwPD have faster reflexive saccadic response time compared to AMC. They also exhibited higher hypermetric saccades and horizontal and vertical fluctuations when focusing on a steady target. The study has also found that there were a large number (nearly 25%) of invalid saccades, i.e., gaze going in the wrong direction, by both groups which were removed automatically before further analysis. The work suggests that recording the gaze using an inexpensive device can have the potential for assisting the neurologists to assess their patients.