Published: July 9th, 2016
A method is described to quantify the quality of visual information processing based on reflexive eye movements in response to specific visual modalities. Reaction times and fixation output parameters are used to characterize visual performance in children with and without visual impairments from 6 months of age.
Visual problems that occur early in life can have major impact on a child's development. Without verbal communication and only based on observational methods, it is difficult to make a quantitative assessment of a child's visual problems. This limits accurate diagnostics in children under the age of 4 years and in children with intellectual disabilities. Here we describe a quantitative method that overcomes these problems. The method uses a remote eye tracker and a four choice preferential looking paradigm to measure eye movement responses to different visual stimuli. The child sits without head support in front of a monitor with integrated infrared cameras. In one of four monitor quadrants a visual stimulus is presented. Each stimulus has a specific visual modality with respect to the background, e.g., form, motion, contrast or color. From the reflexive eye movement responses to these specific visual modalities, output parameters such as reaction times, fixation accuracy and fixation duration are calculated to quantify a child's viewing behavior. With this approach, the quality of visual information processing can be assessed without the use of communication. By comparing results with reference values obtained in typically developing children from 0-12 years, the method provides a characterization of visual information processing in visually impaired children. The quantitative information provided by this method can be advantageous for the field of clinical visual assessment and rehabilitation in multiple ways. The parameter values provide a good basis to: (i) characterize early visual capacities and consequently to enable early interventions; (ii) compare risk groups and follow visual development over time; and (iii), construct an individual visual profile for each child.
The prevalence of brain damage-related visual problems in children has increased. Because visual problems can have great impact on a child's development, early detection in young infants and children at risk is highly important. At present, visual function tests to assess visual sensory functions such as visual acuity and contrast sensitivity (e.g., optotype tests) are applicable in children from 1-2 years of age1. In younger children these tests are based on structured observations of a child's viewing behavior to visual information. The interpretation of such behavior, i.e., by looking at a child's eye movements, can be hampered by oculomotor or attentional dysfunctions of the child, or even by viewing behavior of the observer. Cerebrally mediated visual functions such as visuospatial memory and object recognition are assessed with visual perception tests (e.g., DTVP2). These tests require verbal instructions and communication and can be used from 4-5 years of age. In view of the postnatal development of the visual system and to take advantage of the high level of plasticity early in life, it is desirable to establish the presence and extent of impairments in visual information processing as early as possible. That way, children with (cerebral) visual impairments may maximally benefit from early intervention, visual stimulation, or supportive strategies. Consequently, there is a need for an assessment method of visual information processing that can be used without verbal communication in children and that is based on quantitative results.
Eye movements are a good model to study visually guided orienting behavior to stimuli3,4, and related perceptual and cognitive functions5. Eye movements indicate the focus of visual attention in scenes, and are known to result either from bottom-up (reflexive, salience-driven) or from top-down (intentional, cognitive) processes6. Eye movements are used to direct the fovea, i.e., sharpness of vision, to new objects. The visual content of an object of interest is processed via pathways that run from the retina via the lateral geniculate nucleus to primary visual cortex (V1), and that distribute themselves over cerebral processing areas (e.g., involved in attention, spatial orientation, recognition, memory, and emotions). Eye movements are both a prerequisite for, and a sequel to visual information processing.
Developments in the measurement of eye movements with infrared eye trackers give the possibility to obtain quantitative parameters of oculomotor and visual function. Automated eye trackers are nowadays omnipresent in medical and psychological research involving healthy and clinical populations. Their purpose is not only to study oculomotor function and attention allocation7, but also to answer questions about behavioral and psychological mechanisms8,9. With the rise of accessible and commercial eye tracking systems, they are increasingly used to test vulnerable populations of infants and children10-12, without constraining conditions, complex instructions, or active cooperation12,13. Due to the close coupling of the oculomotor and visual system on an ocular and cerebral level, eye tracking-based methods are pre-eminently suited to assess visual capacities. So far, besides the measurement of visual acuity14, the use of the technique in assessing visual function in children has received relatively little attention.
Our group has combined eye movement measurements with a preferential looking paradigm13. Preferential looking is the preference to fixate patterned surfaces over homogeneous ones15. This principle is applied by using visual stimuli with a target area in one of four quadrants, which differ from the background in terms of one specific visual feature, e.g. coherent form, coherent motion, contrast and color. These visual features are known to be processed by separate peripheral and central visual pathways. For example, information about form is processed by ventral pathways, from V1 to the temporal cortex. Information about motion is processed by dorsal pathways, from V1 to posterior parietal cortex16. Hence, specific stimuli are used to trigger visual information processing in distinct areas of the visual system. If a child is able to see the specific visual information that is presented, that information will attract visual attention in the form of eye movements. These reflexive eye movement responses to the visual stimuli are recorded with a remote infrared eye tracker. That way, eye movement measures provide a communication-free assessment of the quality of various aspects of visual information processing13.
Eye movements provide not only observational data of a child's viewing behavior11, but can also be used for more objective outcome measures. In combination with a carefully designed test paradigm, eye movements can give precise and objective information on visual information processing. This information is obtained by calculating quantitative parameters based on temporal and spatial properties of eye movement responses. Examples of such parameters are reaction time13, fixation time17, saccade metrics7 or cumulative attention allocation18. The availability of these parameters is new to the field of visual assessment in children at a young developmental stage.
The goal of this paper is to present an eye tracking-based method to measure visual information processing in children from the age of 6 months. The measurement set-up and procedure (i.e. nonverbal paradigm, post-calibration, and mobility) specifically apply to using this method in children at risk. A crucial aspect is the analysis of quantitative visual response parameters, i.e. reaction time, fixation duration, and fixation accuracy. These parameters are used to provide reference areas of visually guided responses in typically developing children, to characterize visual information processing in risk groups of children with visual impairments.
The protocol described here was approved by the Medical Ethical Research Committee of the Erasmus Medical Center, Rotterdam, the Netherlands (MEC 2012-097). The procedures adhered to the tenets of the Declaration of Helsinki (2013) for research involving human subjects.
1. Visual Stimuli
Figure 1. Cartoon stimulus. The cartoon stimulus contains various visual modalities (form, motion, color and contrast). This stimulus triggers visual attention, and gives fastest response times in children. Superimposed is an eye movement (gray), going from the lower left corner of the monitor into the target area in the upper right corner (i.e., a reflexive response to the stimulus). Please click here to view a larger version of this figure.
2. Eye Tracking-based Test Paradigm
3. Running the Eye Tracking Experiment
4. Quantitative Analysis of Eye Movements
NOTE: The present protocol is specific to a self-written software program. In order to replicate it, one should write such a software program, e.g., in MATLAB or Python, to quantify the child's visual orienting behavior. In the software program, the following steps are performed for every stimulus type. The present example is focused on Cartoon; the same protocol is applicable to other stimulus types.
Figure 2. Eye movement response to the target area of a stimulus. One eye movement trace (horizontal and vertical directions combined) in distance from the center of the target area (in degrees, y-axis) over stimulus presentation time (in ms, x-axis). The dotted line represents the border of the target area (6° radius). Letters indicate criteria to establish whether the stimulus has been seen: (A) Gaze signal in the first 500 msec; (B) Gaze was not in the target area before 120 msec; (C) Gaze inside the target area for ≥200 msec. Note that in this figure, the depicted presentation time is max 2,000 msec to visualize the first, reflexive response. During testing, total presentation time of all stimuli was 4,000 msec. Please click here to view a larger version of this figure.
|Criterium (Figure 2)
|Verify that the gaze signal:
|Has been recorded for ≥500 msec after stimulus onset
|Capture reflex orienting responses
|Did not enter the target area <120 msec after stimulus onset, and was not already inside the target at the start of stimulus presentation
|Exclude correct performance based on chance
|Was in the target area for ≥200 msec
|Ensure fixation on the target
|Entered the target area within a time window of 1,500 msec, and less than 4 saccades were made
|Exclude visual search behavior
Table 1: Criteria to establish whether a stimulus has been seen. Criteria A, B, and C are visualized in Figure 2.
Figure 3. Visualization of the quantitative parameters RTF, FD, and GFA. One eye movement trace in distance from the center of the target area (in degrees, y-axis) over stimulus presentation time (in msec, x-axis). The vertical red line represents the time at which gaze entered the target area; i.e., Reaction Time to Fixation (RTF). The horizontal red line represents the total time gaze was fixated on the target area; i.e., Fixation Duration (FD). The vertical red arrow represents the width of the fixation trace, in degrees of visual angle, i.e., Gaze Fixation Area (GFA). Please click here to view a larger version of this figure.
The presented method has been applied in two populations of children: a control group of 337 children without visual impairments (mean age (SD) = 4.8 (3.3) years), and a group of 119 children with visual impairments (mean age (SD) = 8.10 (2.96) years) who were recruited at a visual rehabilitation center (Royal Dutch Visio, the Netherlands). Of these children, 74 had ocular visual impairment and 45 had cerebral visual impairments. The results of all control children are visualized in Figures 4-6, separately for reaction time, fixation duration, and gaze fixation area. Reference limits (indicated by black lines) were constructed by fitting a logarithmic function to the control data based on age. These figures serve as a basis for characterizing visual processing functions in children with visual impairments, in terms of impaired or intact function.
The parameter reaction time to fixation (RTF) differentiates between children with- and without visual impairments, and between distinct types of visual impairments. RTF is a measure for the time that is needed to process visual information and execute an eye movement (for calculations refer to a previous study13). The lower the RTF value, the faster the eye movement response. Good repeatability of RTF has been shown in a group of typically developing children from 0-12 years13,21,22, and in children with various types of visual impairments21. Figure 4 shows average RTF to the dynamic Cartoon stimulus over age, for control children, children with cerebral visual impairment (CVI) and children with ocular visual impairment (OVI). RTF values are significantly higher in children with- compared to children without visual impairments (mean difference = 85 msec; t = -13.91, p <0.001, Cohen's d = 1.32) and in children with CVI compared to OVI (mean difference = 99 msec; t = -6.90, p <0.001, Cohen's d = 1.25). These results confirm previously published findings on RTF in subgroups of the present dataset20,24,25.
Figure 4. Average RTF in children with- and without visual impairments. Average RTF values in ms (y-axis) per child, over age (x-axis). Values are shown separately for control children (open circles), children with OVI (black circles), and children with CVI (crosses). The black line represents the upper reference limit of RTF in the control group. RTF values above this line are regarded as deviant, i.e. long reaction times. Please click here to view a larger version of this figure.
Fixation duration is the total amount of time that gaze was fixated within the target area. FD is a measure for sustained visual attention, and is dependent on stimulus presentation time, which is 4 sec in the present example. The parameter fixation duration (FD) also differentiates between children with and without different types of visual impairments. Figure 5 shows mean FD over age, separately for control children, children with CVI, and children with OVI. FD is significantly shorter in children with- than in children without visual impairments (mean difference = 850 msec; t = 11.72, p <0.001, Cohen's d = -1.12), and significantly shorter in children with CVI than in children with OVI (mean difference = 325 msec; t = 2.44, p <0.05, Cohen's d = -0.50). This confirms previous results in children with-, compared to children without visual impairments (Kooiker MJG et al., submitted).
Figure 5. Average FD in children with- and without visual impairments. Average FD values in ms (y-axis) per child, over age (x-axis). Values are shown separately for control children (open circles), children with OVI (black circles), and children with CVI (crosses). The black line represents the lower reference limit of FD in the control group. FD values below this line are regarded as deviant, i.e. short fixation duration. Please click here to view a larger version of this figure.
The parameter gaze fixation area (GFA) is sensitive to detect disturbances in oculomotor control, in particular nystagmus. GFA represents the size of the area of fixation in degrees, and is a measure for fixation accuracy (for calculations see previous studies13,23). A small area of fixation indicates high fixation accuracy. GFA depends on the size of the stimulus and the corresponding target area (i.e., a 6º radius in the present example). Good repeatability of GFA has been shown in a group of typically developing children from 0-12 years13, 21, and in children with various types of visual impairments21. Figure 6 shows mean GFA in response to the cartoon stimulus over age, separately for control children, children with the oculomotor impairment nystagmus, and children with visual impairments but without nystagmus. GFA values are significantly larger, i.e. lower fixation accuracy, in children with- compared to children without visual impairments (mean difference = 1.34º; t = -25.09, p <0.001, Cohen's d = 2.37). In addition, children with nystagmus have lower fixation accuracy than children without nystagmus but with other types of visual impairment (mean difference = 0.71º; t = 5.03, p <0.001; Cohen's d = 1.04). This is consistent with previously published findings on GFA in subgroups of the present dataset20,24,25.
Figure 6. Average GFA in children with and without visual impairments. Average GFA values in degrees (y-axis) per child, over age (x-axis). Values are shown separately for control children (open circles), children with visual impairment and nystagmus (asterisk), and children with visual impairment without nystagmus (black diamond). The black line represents the upper reference limit of GFA in the control group. GFA values above this line are regarded as deviant, i.e. low fixation accuracy. Please click here to view a larger version of this figure.
The presented measurement set-up combined with quantitative eye movement analysis provides a distinct characterization of visual processing functions in various groups of children with oculomotor and visual impairments. The key feature of this paradigm is that performance is based on eye movement responses to visual stimuli that are triggered in a reflexive manner. No specific verbal instructions are given and there is no need for children to verbally respond. The parameters RTF, GFA and FD show significant differences between groups of typically developing- and visually impaired children, despite the limited spread of parameter values that exists in each group (Figures 4-6). Thus, depending on the evaluated parameter, some typically developing children can show deviant performance, while some children with visual impairments show 'normal' performance. Ultimately, multiple outcome measures in response to multiple visual modalities should be considered on an individual level. A summary of all outcome measures provides a unique characterization of visual information processing capacities, which may be converted in a visual profile in children from 6 months of age.
Several studies have shown the value of remote eye tracking in vulnerable populations of children, to infer attentional or psychological capacities9,12,18. Whereas most studies rely on behavioral observations and the use of instructions, a distinct feature of the current paradigm is the nonverbal, quantitative approach. Critical steps within the protocol therefore include the stimuli that are based on preferential looking, the mobile measurement set-up, and the custom calibration and analysis software. The presented extension of observation-based results with elaborate analysis methods provides standardized and detailed results on visual processing functions. This is in line with work on the assessment of infant visual acuity with an eye tracker14, and work on gaze control in various disorders7. The method is flexible and enables mobile assessment which is indispensable when performing clinical assessments in young children or children with multiple disabilities. Therefore, it is suited to measure oculomotor and visual processing capacities in virtually all children that are capable of watching a monitor.
The significance of this method with respect to existing visual diagnostic methods (i.e., validity) has been studied as a first step toward clinical implementation. The present paradigm was combined with currently used visual function assessment (VFA) in children. Observations of oculomotor and visual functions that are based on eye movement recordings were comparable with standard behavioral observations of these functions. Moreover, eye tracking parameters, e.g., fixation duration and saccadic direction, provided additional value in characterizing oculomotor and visual performance in children during VFA (Kooiker MJG et. al., 2015, submitted). The major gain of the presented method lies in the possibility to assess more visual functions than is currently done in visual function assessments at a young age, and to assess them in a quantitative manner26. A limitation with respect to existing methods is that, without adaptations, it is not yet possible to thoroughly assess visual acuity or visual field with the present test battery14.
Although we limited ourselves to the presentation of results from cartoon stimuli, in future applications different visual modalities can be tested using other stimuli (e.g., distinct forms, motion, color and contrast information)22,20,25. That way, specific visual processing areas beyond the primary visual pathways are targeted, such as visual association areas in temporal or parietal cortex. A limitation of the method is that the present visual stimuli merely trigger the detection of visual input, and invoke the initial stage of visual processing. These stimuli do not target higher-order functions that become relevant after stimulus detection and that are normally measured with visual perception tests. Although their execution without the use of communication is challenging, an eye tracking-based paradigm is a promising future format for the detection of perception-related information, e.g. visual search, -memory or -selective attention.
In sum, detailed eye movement responses to various types of visual stimulation provide a comprehensive characterization of visual information processing functions, early in development. Consequently, for each child an individual visual profile in terms of intact and impaired functions can be created. Such a profile may provide detailed information about strengths and weaknesses in oculomotor and visual function. It may be used as a starting point for support in daily life, and for teacher and caregiver education. The quantitative information that has become available with this method can be advantageous for following visual development over time, and for monitoring visual interventions and rehabilitation programs.
The authors declare that they have no competing financial interest.
The authors thank daycare centers (Wasko, Alblasserwaard) for their support in recruiting the control group, and Mark Vonk for his help in data collection in the control group. The authors also thank the children from the control group and the children who are clients from Royal Dutch Visio for participation in the study. The authors are grateful to the children and their parents for participation in the video.
The development of the method was supported by a grant from the Novum Foundation: a non-profit organization providing financial support to (research) projects that improve the quality of life of individuals with a visual impairment (www.stichtingnovum.org). Financial support for the current study was provided by 'ZonMw Inzicht' (Netherlands Organization for Health Research and Development-Insight Society), grant number: 60-00635-98-10.
|Tobii T60 XL
|Tobii Technology: http://www.tobii.com
|remote infrared eye tracker
|Tobii Technology: http://www.tobii.com
|eye tracker software
|data analysis software
Copyright © 2024 MyJoVE Corporation. All rights reserved