A new Thomas RECORDING brand
Humans move their eyes approximately 2 – 3 times per second, significantly more often than their heart beats. Already in 1908, Diefendorf & Dodge described characteristic malfunctions of eye movements in patients with neurological and psychiatric diseases. Since then, the lists of affected eye movements, diseases and corresponding brain regions have been extended massively (see Leigh, R. J. & Zee, 2015).
For patients with Parkinson’s disease, recent basic research identified several particularly suitable eye movements that are used as biomarkers by our medical TOM (Thomas Oculus Motus) eye tracking systems to support the diagnosis of the disease and to be able to objectively monitor the course of therapy. In contrast to conventional clinical-neurological examinations based on an evaluation of the symptoms, the experience of the neurologist and complex imaging procedures as method of elimination of other potential diagnosis, which can cause a high risk for misdiagnosis and delayed diagnosis (Beach & Adler, 2018), eye movement parameters provide clear and objective diagnostic support. Furthermore, certain eye movement parameters correlate well with the typical symptom rating scales like Hoehn & Yahr or UPDRS, as well as the individual medication of a patient, allowing for a more detailed classification and monitoring of the disease state and therapy progress control.
The medical TOM eye tracking systems (TOM mobile and the TOM stationary) present visual stimuli to the patient and track the resulting eye movements. The TOM systems use integrated motion-tracking cameras and cloud based artificial intelligence (AI) to quickly analyze patient´s eye movements. The AI compares the data of the patient’s eye movements with a large database of eye movement patterns from healthy control subjects and other patients suffering from PD to identify characteristic biomarkers associated with Parkinson´s disease. Some of the test paradigms used by our medical TOM systems are described below.
1. A mixture of Pro- & Anti-Saccades
A Pro-Saccade is a fast, ballistic eye movement towards an emerging visual target in the periphery. An Anti-Saccade, on the other hand, requires the subject to suppress the reflexive eye movement towards the emerging target and instead plan and execute a gaze shift towards an imaginary target, mirrored at the vertical meridian (Figure 1).
Figure 1: Schematic representation of a Pro- (left) and Anti-Saccade Task (right). A subject has to fixate the central fixation point, which color will indicate if a pro- or an anti-saccade is requested. After a couple of seconds, a second target will appear in the periphery randomly left or right of the center. The subject is instructed to perform the indicated eye movement towards or away from the target as fast as possible after its appearance.
There are numerous studies underpinning the clinical relevance of this task for neurological and psychiatric diseases in general (e.g. Everling & Fischer, 1998) and Morbus Parkinson in particular (Waldthaler et al., 2021). A variety of saccade related parameters, e.g. error rate, latency or gain, have been shown to qualify as biomarkers for patients with Parkinson’s disease and moreover correlate with established rating scales and medication dose (Waldthaler et al. 2019).
Figure 2: Selection of saccade eye movement parameters and their meaning for the diagnosis, classification and medication evaluation of patients with Parkinson’s disease. A Visually guided saccade (VGS) gain of healthy controls (HC) and patients with Parkinson’s disease (PD) in Hoehn & Yahr Stage 2 (H&Y2) or Stage 3 (H&Y3) and OFF medication state. * indicate a significant group difference (ANOVA) compared to healthy controls. B Linear regressions of vertical gain and UPDRS (upgain: r = −0.43; p = 0.01; downgain: −0.38; p = 0.02). C Saccadic latency of HC and PD patients in OFF and ON medication state. The brackets show the results of the paired t-tests comparing OFF and ON state. D Anti-saccade (AS) error rates of HC and PD patients in H&Y2 and H&Y3 in OFF and ON medication state. The AS error rate was increased only in H&Y3 patients (ANOVA, compared to HC; * shown without brackets) and improved after levodopa in these patients (paired t test, shown with brackets). (*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.005; error bars show standard deviation) Adapted from Waldthaler et al. (2019)
2. Smooth Pursuit Eye Movements
Smooth Pursuit Eye Movements are performed to keep gaze on a moving object, like the dot below:
Figure 3: Example stimulus to elicit horizontal & vertical Smooth Pursuit Eye Movements in a subject.
Smooth Pursuit has been shown to be affected in patients with Parkinson’s disease (e.g. Gorges et al., 2014; Zhang et al., 2018). This is expressed by the eyes having trouble to keep up with the movement of the target and lagging behind it. This lag triggers frequent intrusions of the otherwise smooth trajectory by catch-up saccades, which realigns the gaze and the target position (Figure 4).
Figure 4: Horizontal smooth pursuit eye movements elicited by a sinusoidal oscillating spot (𝑓 = 0.125 Hz) and exemplified for a representative age-matched healthy control (upper panel) and a Parkinson’s disease patient (PD, lower panel). The PD patient presents with severely affected SPEM, frequently interrupted by catch-up saccades. Although SPEM is heavily impaired in PD, patients retain the ability to perform episodes of genuine smooth pursuit (arrow). Adapted from Gorges et al. (2014)
The lag of the eye is quantified via the Smooth Pursuit Gain (eye velocity / target velocity during the genuine smooth pursuit episodes), or Smooth Pursuit velocity accuracy (percentage of time during the entire movement range when the eye movement velocity was within the target velocity boundaries of less than 20% absolute error from the visual target velocity), which both can be used as a biomarker. Furthermore, it has been shown that therapeutic measures like deep brain stimulation (DBS) can significantly improve the Smooth Pursuit Gain and velocity accuracy in patients with Parkinson’s disease (Figure 5; Nilsson et al., 2013), providing an additional use of these parameters for therapy monitoring.
Figure 5: A Mean gain values (eye velocity / target velocity) of 25 patients with Parkinson’s disease for four target velocities with DBS OFF and ON. A value below 1.00 represent that the average smooth pursuit velocity was below the target velocity and thus the eye lagged behind the target. B Mean Smooth Pursuit velocity accuracy values with DBS OFF and ON. A value of 100 % represents that the eye movement velocity was always within the boundaries around the target velocity. (# denotes P < 0.1; * denotes P < 0.05; ** denotes P < 0.01 and *** denotes P < 0.001). Adapted from Nilsson et al. (2013)
3. A Free Viewing Task
In a Free Viewing Task, subjects just have to view images or videos and freely explore them for a certain period of time. It has been shown that this visual exploration is altered in patients with Parkinson’s disease (e.g. Archibald et al., 2013; Zhang et al., 2018) leading to characteristic scan paths that are distinct from those of healthy control subjects (e.g. Figure 6). The exploration strategy can be quantified by countless eye movement parameters, such as the size of the exploration area (colored areas), the number of fixations or the dwell time at specific salient locations.
Figure 6: Representative visual scan paths and eye movement parameters during a Free Viewing Task performed by patients with Parkinson’s disease (PD) and age-matched healthy control subjects (Control). E Example of eye traces in one representative PD patient F Example eye traces of one representative healthy control subject. G Boxplot of saccade frequency during Free Viewing across all 37 PD patients and 39 control subjects. H Boxplot of saccade frequency during Free Viewing across all 37 PD patients and 39 control subjects. *p < 0.05; **p < 0.01; ***p < 0.001.
A recent study by Tseng et al. (2012) used 224 eye movement features acquired during 15 minutes of watching TV to correctly classify patients with Parkinson’s disease from health subjects with an accuracy of almost 90%.
Finally, the stimuli of the TOM systems are designed to maximize the efficiency in eliciting the desired biomarkers in the patients and thereby optimizing the duration of examination and finally patient compliance.
Archibald, N. K., Hutton, S. B., Clarke, M. P., Mosimann, U. P., & Burn, D. J. (2013). Visual exploration in Parkinson’s disease and Parkinson’s disease dementia. Brain, 136(3), 739-750.
Beach, T. G., & Adler, C. H. (2018). Importance of low diagnostic accuracy for early Parkinson’s disease. Movement Disorders, 33(10), 1551-1554.
Everling, S., & Fischer, B. (1998). The antisaccade: a review of basic research and clinical studies. Neuropsychologia, 36(9), 885-899.
Gorges, M., Pinkhardt, E. H., & Kassubek, J. (2014). Alterations of eye movement control in neurodegenerative movement disorders. Journal of ophthalmology, 2014.
Nilsson, M. H., Patel, M., Rehncrona, S., Magnusson, M., & Fransson, P. A. (2013). Subthalamic deep brain stimulation improves smooth pursuit and saccade performance in patients with Parkinson’s disease. Journal of neuroengineering and rehabilitation, 10(1), 1-12.
Tseng, P. H., Cameron, I. G., Pari, G., Reynolds, J. N., Munoz, D. P., & Itti, L. (2013). High-throughput classification of clinical populations from natural viewing eye movements. Journal of neurology, 260(1), 275-284.
Waldthaler, J., Tsitsi, P., & Svenningsson, P. (2019). Vertical saccades and antisaccades: complementary markers for motor and cognitive impairment in Parkinson’s disease. NPJ Parkinson’s disease, 5(1), 1-6.
Waldthaler, J., Stock, L., Student, J., Sommerkorn, J., Dowiasch, S., & Timmermann, L. (2021). Antisaccades in Parkinson’s Disease: A Meta-Analysis. Neuropsychology Review, 1-15.
Zhang, Y., Yan, A., Liu, B., Wan, Y., Zhao, Y., Liu, Y., … & Liu, Z. (2018). Oculomotor Performances Are Associated With Motor and Non-motor Symptoms in Parkinson’s Disease. Frontiers in neurology, 9, 960.
SIGN UP TO OUR NEWSLETTER