Eye movement during reading in young adults with cerebral palsy measured with eye tracking.

Postgraduate Medicine

ISSN: 0032-5481 (Print) 1941-9260 (Online) Journal homepage: http://www.tandfonline.com/loi/ipgm20

Eye Movement during Reading in Young Adults with Cerebral Palsy Measured with Eye Tracking Renée Lampe Prof. Dr., MD, Varvara Turova PhD, Tobias Blumenstein MSc & Ana Alves-Pinto PhD To cite this article: Renée Lampe Prof. Dr., MD, Varvara Turova PhD, Tobias Blumenstein MSc & Ana Alves-Pinto PhD (2014) Eye Movement during Reading in Young Adults with Cerebral Palsy Measured with Eye Tracking, Postgraduate Medicine, 126:5, 146-158 To link to this article: http://dx.doi.org/10.3810/pgm.2014.09.2809

Published online: 13 Mar 2015.

Submit your article to this journal

Article views: 3

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=ipgm20 Download by: [University of Pennsylvania]

Date: 12 September 2015, At: 08:23

C L I N I C A L F E AT U R E S

Eye Movement During Reading in Young Adults With Cerebral Palsy Measured With Eye Tracking

Downloaded by [University of Pennsylvania] at 08:23 12 September 2015

DOI: 10.3810/pgm.2014.09.2809

Abstract Renée Lampe, Prof. Dr. MD Varvara Turova, PhD Tobias Blumenstein, MSc Ana Alves-Pinto, PhD Klinikum rechts der Isar, Technische Universität München, Munich, Germany

Background: Cerebral palsy is a nonprogressive brain disorder associated with lifelong motor impairments and often with cognitive deficits, impaired communication, and impaired sensory perception. Vision deficits, in particular, occur frequently in cerebral palsy and can lead to reading difficulties. Objective: Investigate the extent to which the motor impairments in this clinical group affect patients’ ability to read. Methods: An eye-tracking system was used to record the eye movements during a reading task in 31 adults diagnosed with cerebral palsy and in 10 healthy controls. Participants were asked to read out loud 1 to 5 excerpts from children’s books. Results: In comparison to the healthy readers, cerebral palsy patients took longer to read the excerpts; made more saccades, fixations, and regressions; and made shorter saccades. Average fixation times were similar between the 2 groups, but the average saccade duration was significantly longer for the cerebral palsy group, as a function of the degree of severity of motor impairment. The latter was not a determinant of the level of text comprehension achieved by these patients. Conclusions: Objective measures of eye movement during a reading task can be obtained in cerebral palsy patients using eye-tracking techniques. Results suggest that cerebral palsied patients may experience difficulties in searching for words during reading. Keywords: cerebral palsy; eye tracking; vision; reading; Gross Motor Function Classification System

Introduction

Correspondence: Renée Lampe, MD, Klinikum rechts der Isar, Technische Universität München, Ismaningerstraße 22, 81675 Munich, Germany. Tel: +49-(0)89-4140-6360 E-mail: [email protected]

146

Cerebral palsy (CP) is a nonprogressive brain disorder caused by damage to the immature developing brain.1 The etiology may be diverse, and includes lack of oxygen, an infection during the mother’s pregnancy, blood group incompatibility, genetic factors, and other noxious agents. A statistically significant correlation between preterm birth and cerebral palsy has also been reported.2 Nevertheless, in some cases the cause remains unclear.3 The range of disorders that can arise with CP is wide. Depending on the extent and topography of the damage, in addition to motor impairments, deficits in perception, reading, language, and cognition can occur, leading to limitations in daily activities. In particular, visual impairments are frequent in CP, affecting almost 50% of children with CP.4 These impairments can have their origin in either the eye or the brain. Diseases that affect the eye can lead to the development of retinal damage, reduced visual acuity, or refractive errors. Fazzi et al5 examined 129 children and adolescents with CP with the aim of describing visual dysfunctions in this population and assessing whether different types of CP are related to distinct visual deficits. They found that

© Postgraduate Medicine, Volume 126, Issue 5, September 2014, ISSN – 0032-5481, e-ISSN – 1941-9260 ResearchSHARE®: www.research-share.com • Permissions: [email protected] • Reprints: [email protected] Warning: No duplication rights exist for this journal. Only JTE Multimedia, LLC holds rights to this publication. Please contact the publisher directly with any queries.


Eye Movement During Reading in Young Adults With Cerebral Palsy

the visual dysfunctions in diplegia are mostly deficits related to refractive errors, strabismus, and abnormal saccade movements. Patients with hemiplegia were found to have strabismus and refractive errors. Children with quadriplegia demonstrated a severe neurological and ophthalmological profile of ocular abnormalities, oculomotor dysfunction, and decreased visual acuity. Vision impairments that originate in the brain, designated as central visual perception disorders,6 can lead to perceptual and visuomotor deficits. Problems in gaze coordination, with difficulties initiating and performing saccades, unstable fixations, and difficulties in scanning the environment have been observed in children with cerebral visual impairment of hypoxic-ischemic origin.7 Analysis of eye saccades with electro-oculography in children with CP found slow saccades and frequent corrective saccades.8 Deficits in visuomotor coordination and difficulties in finding targets in a visual search task have also been described in premature children who otherwise had performed normally on standard screening tests.9 Visual-attentional deficits have also been suggested to underlie the poorer performance of children with CP in refixating in a central visual target when a competing stimulus is simultaneously presented.10 Deficits in visual functions lead to underperformance in functional tasks: in reading,11,12 in visual-perception tasks,13 as well as in hand coordination tasks.14 Kozeis et al11 used the Developmental Eye Movement (DEM) test15,16 to investigate microsaccadic movement during reading in children with CP and in a control group. The DEM is a visual-verbal test of oculomotor skill. In this study 20.95% of the children with CP were reported to present with pure oculomotor problems, 32.38% presented with a visual perceptual problem, and 27.61% presented with a combination of oculomotor and perceptual problems.15 Exactly what these alterations in eye movement can tell us about the reading abilities in patients with CP requires further investigation. The DEM may not provide the most accurate measure of eye movement,17 and it is unclear how much the control of eye movement while reading is carried out by low-level versus high-level processes.18 Nevertheless, less skillful readers (eg, preschool children) typically show longer fixations and shorter saccades and make more fixations and regressions than skillful readers (eg, schoolchildren).19 In a preliminary study Lampe et al20 evaluated the possibility of using mobile eye tracking glasses to investigate perceptual disorders in children and adolescents with CP. A higher number of regressions with typical repetitions of reading the same syllables were observed in these children,

consistent with observations reported for less skillful readers. In the present study, the eye tracking technique was used to evaluate quantitatively changes in eye movement during reading in people who had sustained brain damage as infants, and to investigate the dependency of reading-related eye movements on the severity of disability, as measured by the Gross Motor Function Classification System (GMFCS) level.21 The GMFCS is an internationally recognized classification system that assesses motor function and assigns 1 of 5 levels, with higher levels indicating more severe motor impairment. It is helpful not only for classification of motor function but also in assessing motor skill development over time. In the current study, the classification was performed by an orthopedic surgeon or a physical therapist. Children with more severe sensorimotor deficits (highest GMFCS levels) are more likely to develop diverse visual deficits,22 such as myopia, absence of binocular fusion, dyskinetic strabismus, severe gaze dysfunction, and cerebral visual impairment. Children with milder motor impairments (levels 1 and 2) present sensorimotor deficits similar to those of neurologically normal children with strabismus and amblyopia. One of the aims of the current study, therefore, was to assess the extent to which the severity of motor dysfunction is reflected in different degrees of altered eye movement during reading. Better understanding of eye movement peculiarities in patients with CP will help in adapting therapies to their needs and in optimizing the configuration of the patients’ workplace. Eye tracking provides better insight into patients’ text understanding abilities, and therefore can guide the way reading material is presented/displayed to these patients (eg, by styling the text using shorter words and separating the text into several shorter parts).

Method Participants

The test group was composed of 31 adults diagnosed with CP; their ages ranged from 19 to 57 years. The group included 17 men and 14 women. Eligibility criteria included a diagnosis of CP and the ability to read. A control group of 10 healthy subjects (9 women and 1 man), ranging in age from 27 to 36 years, was also included in the study. This group was composed of personnel working as therapists in the center for people with physical disabilities from which the patients with CP were recruited. Participation in the study was voluntary, and the subjects were informed that they could withdraw from the study at any time without giving a reason. All procedures were approved by the ethics committee of the Technische Universität München.

© Postgraduate Medicine, Volume 126, Issue 5, September 2014, ISSN – 0032-5481, e-ISSN – 1941-9260 147 ResearchSHARE®: www.research-share.com • Permissions: [email protected] • Reprints: [email protected] Warning: No duplication rights exist for this journal. Only JTE Multimedia, LLC holds rights to this publication. Please contact the publisher directly with any queries.

Lampe et al


Clinical Assessment

The patients’ IQ level had been determined in earlier assessments when the patients first came to the center. The IQ testing was done with the Hamburg-Wechsler Intelligence Test for Children, which is a German-language version of the Wechsler Intelligence Scale for Children. The children’s test was used because the Wechsler Adult Intelligence Scale would have been too sophisticated for these patients. The average IQ of the patient group was 73.2, ranging from 58 to 89. IQ values in the range of 70 to 84 indicate a learning disorder. (The control group was not tested with the HAWIK, and it was assumed that the control subjects had IQs in the normal range.) Most of the patients’ CP was caused by a premature birth, but in 3 patients it was caused by a posttraumatic injury in early childhood. In most cases, a brain hemorrhage that was found after birth was noted in the clinical reports. According to some medical reports and parents’ statements, radiological findings included periventricular leukomalacia and enlargements of the ventricles cystic defects. However, no radiological images were available for verification. Patients’ motor impairment level was assessed with the GMFCS classification system,21 as described above. Patients with CP often have dysarthria—difficulty in articulating words—which reflects the underlying spastic muscle tone of the muscles involved in articulation. To control for the influence of these difficulties during text reading, a diagnostic assessment of reading sense and speech disorder was carried out for all patients by a speech therapist. Dysarthria was identified in 15 of the 31 patients (Table 1). All patients in this study were under regular ophthalmological management. Refractive errors were corrected with glasses. The current diopters of the glasses were documented. Some pathological findings had been successfully treated during childhood; 42% of patients wore glasses for vision correction because of myopia or hyperopia. Most of the participants visited a center for persons with physical disabilities since beginning school or kindergarten, where special eye training programs were implemented and medical recommendations were strictly followed. Teachers were instructed to apply bandages to the patients’ eyes, one eye at a time, to prevent the eyes from becoming tired. They also made sure the children wore their glasses regularly to compensate for the strabismus. No cases of optic atrophy or glaucoma were identified.

Eye Tracking (Gaze Tracking System)

The fixed eye tracking system EyeFollower (Interactive Minds GmbH, Dresden, Germany) was used to record 148

the eye movements (Figure 1). This remote system uses 4 high-resolution cameras with telephoto lenses integrated in the desktop that track the head position and adjust the eyetracking cameras to the subject’s eyes. After a successful 5-point calibration, the eye cameras can automatically follow the movements of both eyes during reading in real time, with a recording rate of 120 Hz. Data was recorded with the software Nyan (Interactive Minds), and analyzed in MATLAB. The system comprises a 24-inch screen on which the texts were displayed. The participants sat 60 to 80 cm away from the screen, which enabled them to comfortably touch the screen (Figure 1). While the patient was reading the text out loud, 4 high-resolution cameras inserted in the lower part of the setup followed the patient’s eye movements. All eye-tracking measurements were made in the same room at the center.

Text Reading Task

Five different texts were selected for the reading task. They consisted of excerpts of stories for children, so the vocabulary and style, similar across all texts, were accessible for all patients regardless of their reading habits and reading skills. The texts differed in their length and in the words used, with some having words that were more difficult to articulate and some giving information in a more indirect way than others. The same sans serif font was used in all experimental sessions and for all subjects. All the participants were native German speakers. The participants sat in front of the computer screen and were asked to read out loud the text shown on the screen (Figure 1). Each text was read only once. After reading the text, patients completed a 10-item questionnaire together with the therapist to assess their text comprehension level. The test took place in a quiet room, and all participants were asked the same questions.

Data Analysis

The eye tracking system enables the recording and analysis of various parameters such as the mean duration of fixations, the saccade distance and duration, the accuracy of saccades, and the pupil size. Figure 2 provides an example of the eye-tracking data obtained during a reading exercise. The circles illustrate fixation points, with larger circles reflecting longer fixations times. Fixation points are correctly located over words. The line illustrates the saccade movement between fixation points. The line follows a logical route from left to right, then moves toward the left to the beginning of the following line, and then moves toward the right as the new line is being read, and so




Table 1. Participants’ Demographical Data and Patients’ Clinical Information Patient number

Sex

Age

GMFCS level

Diagnosis

Dysarthria

Glasses

p1 p2 p3 p4 p5 p6 p7 p8 p9 p10 p11 p12 p13 p14 p15 p16 p17 p18 p19 p20 p21 p22 p23 p24 p25 p26 p27 p28 p29 p30 p31 h1 h2 h3 h4 h5 h6 h7 h8 h9

M F M F F M M F M F F M M F M F F M M F F M F M F M F M M M M F F F F F F F F M

46 46 55 29 41 45 31 31 35 50 49 28 46 24 34 22 39 20 26 24 57 48 49 32 32 26 33 23 22 19 20 30 36 29 29 31 27 36 28 32

4 4 4 4 3 3 4 4 3 4 3 4 4 1 4 1 4 1 4 5 4 5 3 4 1 4 4 4 3 4 1

Bilateral dyskinetic CP Bilateral spastic CP Bilateral spastic CP Bilateral spastic CP Bilateral spastic CP Bilateral spastic CP Bilateral spastic CP Bilateral spastic CP Bilateral spastic CP Bilateral dyskinetic CP Bilateral spastic CP Bilateral spastic CP Bilateral spastic CP Unspecific CP Bilateral spastic CP Unilateral spastic CP Bilateral spastic CP Unspecific CP Bilateral spastic CP Bilateral spastic CP Bilateral spastic CP Bilateral spastic CP Bilateral spastic CP Bilateral spastic CP Ataxia Bilateral spastic CP Bilateral spastic CP Bilateral spastic CP Leg-dominated bilateral spastic CP Bilateral spastic CP Unilateral spastic CP Healthy Healthy Healthy Healthy Healthy Healthy Healthy Healthy Healthy

No No Yes Yes No No Yes Yes No No No No Yes No No No No No No Yes Yes Yes Yes Yes No Yes Yes Yes Yes Yes No

Yes Yes Yes No Yes Yes Yes No Yes Yes Yes Yes Yes No No No No No Yes No Yes No Yes No No No No No No No No No No No No No No No No No

Abbreviations: CP, cerebral palsy; GMFCS, Gross Motor Function Classification System.

forth until it reaches the end of the text on the right side of the final line. Figure 2B presents more fixation points as well as longer fixation points than Figure 2A. This, however, constitutes an extreme case, with unusually long fixations (also within the CP group). Due to a mistake in the calibration, the top line of the text is not detected by the eye tracking. Probably the reader has changed position with respect to the screen. This has led to an incorrect reconstruction of the view point. Unfortunately, the calibration file cannot be recovered and the reading path corrected. Nevertheless, the patient read the text out loud completely and correctly.

Eye-tracking recording was disturbed on some occasions by reflection in the patient’s glasses and movements of the head, which affected the accuracy of the pupil position. This led to an incorrect matching between the location of the word that was being read and the simultaneous fixation point. This type of error could only be detected during the off-line analysis of the recorded data. Data sets affected by these errors were excluded from further analysis. A total of 60% of the initial recorded data sets were considered valid and were analyzed further. Not all texts could be completely read by all patients, with some texts being too long or containing words that


Lampe et al


Figure 1. Experimental setup used in the reading tests and eye-tracking recording.

were too difficult to articulate. The results presented below are based on the analysis of the eye-tracking data that were considered valid after off-line analysis and for the texts that were read to the end.

Results

The eye tracking process shown in Figure 2A presents an example of a healthy reader: the eye moves from word to word from the left to the right and from the end of the line to

the beginning of the next line. The maximum fixation times were in the range of 700 to 800 ms. Figure 2B illustrates an example of reading by a patient with CP, a GMFCS level of 4, and IQ in the “dull-normal” range. This patient constitutes an extreme case, with extreme long durations of fixations with maximum values in the range of 2.1 to 2.2 seconds. Three different reading patterns of this text are shown in Figure 3, one by a healthy patient (Figure 3A), another by a unilateral palsy patient (Figure 3B), and another by a bilateral CP patient with dysarthria (Figure 3C). The horizontal axis shows the position of the viewpoint during reading (in arbitrary units); a value of 0 indicates the left border of the display. Time (in seconds) is represented in the vertical axis starting at the top and progressing downward. As the participant reads through a line, the eye viewpoint moves from left to right, and as the reader moves through the text the eye viewpoint moves from the top to the bottom of the screen. Figure 3A illustrates the reading pattern typical of a good reader, with a series of saccades forward toward the right and small saccades backward and a few longer backward saccades as the viewpoint moves to the beginning of the following line. Both CP patients (Figure 3B,C) made more and shorter saccades than the healthy reader, especially the bilateral CP patient as shown in the inset of

Figure 2. Two examples of eye-tracking data obtained during text reading: (A) for a healthy reader; (B) for a patient with CP.

Abbreviation: CP, cerebral palsy.

150




Figure 3. Examples of eye viewpoint position collected with the eye-tracking system for 3 participants during the reading of text 1: (A) healthy participant, (B) CP patient without dysarthria, (C) CP patient with dysarthria. (C, inset): Enlarged view of the upper part of the plot in C illustrates the numerous small regressions made by this patient.


Figure 3C. The higher number of fixations and longer scan path translated into a longer time required to read the text. More regressive saccades could also be observed for both CP readers. All these indications confirm the difficulties in performing the reading task by the CP patients, who require

more processing (more fixations) and who need to reread the text (more regressions). The frequency of the types of saccades (long, short, forward, backward) for each reader can be described in a histogram of the saccades’ path length during the reading



Lampe et al

task (Figure 4). Long saccades are more prevalent in a person with a regular reading behavior. Though the scan path includes jumps, the reader can capture the text or the word content fast. This behavior of a healthy participant (Figures 3A and 4) generates a bimodal distribution of the saccade’s length that clearly differs from the patterns of the CP fast and slow readers, which show unimodal skewed distributions, particularly for the bilateral CP slow reader, with a higher number of shorter saccades and a clear spread toward shorter and negative (regressions) saccades. In the examples shown, the patients made 333 and 140 total saccades as compared with against only 114 saccades made by the healthy reader. The median regression length was also shorter for the CP readers (median –100) compared with the healthy reader (median –150). The unilateral CP fast reader presents an intermediate case between the healthy and the bilaterally affected CP slow reader. As illustrated in Figure 3, the scan path and reading time, derived from the eye-tracking data, were longer for the CP patients. The longer reading time was observed also for all texts for which valid eye tracking data could be collected. Reading rates are illustrated in Figure 5. A lower reading rate indicates a longer reading time. To increase the statistical power of the comparison between groups of participants, and because the reading rate did not differ significantly between different texts (2-way ANOVA with 3 levels for the participant group factor23 [healthy, CP with no dysarthria, CP with dysarthria] and 5 for the text factor, F[4,60] = 2.02, P = 0.1029), the results are presented without differentiating between texts. The same procedure will be applied to the other variables below, as the text type had no effect on the ratio of average fixation duration to average saccade duration Figure 4. Histogram of the saccade displacement (in arbitrary units) for a healthy participant, a unilateral CP patient, and a bilateral CP patient. Positive values indicate eye movement from left to right and negative values indicate movements in the opposite direction (regressions).

Abbreviation: CP, cerebral palsy.

152

Figure 5. (A) Reading rate (number of words per minute) for the healthy individuals (dots), CP patients with no dysarthria (circles) and CP patients with dysarthria (crosses) CP patients. Asterisks and bars indicate the mean value for the group of nearby symbols and the associated standard deviation, respectively. (B) Reading rate as a function of GMFCS level.


(2-way ANOVA with 3 levels for the participant group factor and 5 for the text factor, F[4,59] = 0.43, P = 0.7844); the average duration of fixations (2-way ANOVA with 3 levels for the group factor and 5 levels for the text factor, F[4,60] = 0.44, P = 0.7821); and the average duration of saccades (2-way ANOVA with 3 levels for group factor and 5 levels for text factor, F[4,59] = 0.73, P = 0.5738). The reading rate was lower for the patients with dysarthria (individual results represented by “ × ” symbols in Figure 5, with a mean of 49 words per minute compared with 134 words per minute for the healthy group; mean values are depicted by asterisks) but also for patients not diagnosed with




dysarthria (“o” symbols in Figure 5, mean of 94 words per minute). Differences between healthy and CP with no dysarthria groups were statistically significant (Wilcoxon rank sum test corrected for multiple comparisons, W = 1032, z = 4.53, P , 0.001, r = 0.62) as well as between the healthy and CP with dysarthria groups (Wilcoxon rank sum test corrected for multiple comparisons, W = 742, z = 4.94, P , 0.001, r = 0.78) and between the CP with no dysarthria and CP with dysarthria groups (Wilcoxon rank sum test, W = 122, z = –3.50, P , 0.001, r = –0.57). Wilcoxon rank sum tests, corrected for multiple comparisons, showed statistically significant differences in reading rate between GMFCS levels 1 and 3 (Ws = 281, z = 3.06, P = 0.002, r = 0.62), GMFCS levels 1 and 4 (Ws = 397, z = 4.39, P , 0.001, r = 0.80), and GMFCS levels 1 and 5 (Ws = 247, z = 2.68, P , 0.007, r = 0.57). No differences were obtained among the groups with GMFCS levels 3, 4 and 5. The slower reading times in the patient groups may be due to difficulties in word articulation but can also result from longer processing times (ie, longer fixation durations) or longer searching times (ie, longer saccade duration). Investigation of possible underlying causes for the longer reading times focused on the analysis of the average fixation duration and the average saccade duration separately. Values of these 2 variables are represented in Figures 6 and 7, respectively. The average fixation time was similar for all 3 groups (healthy, CP with no dysarthria, CP with dysarthria; Figure 6A), with no statistically significant differences after the Wilcoxon rank sum test corrected for multiple comparisons. No differences among the GMFCS levels were found (Figure 6B; 1-way ANOVA with 5 levels for the GMFCS level factor, F[4,62] = 0.72, P = 0.5832). The duration of the saccade, however, was significantly longer for the CP group diagnosed with dysarthria (Figure 7A) in comparison with both the healthy group (Wilcoxon rank sum test, Ws = 408, z = –4.88, P , 0.001, r = –0.77) as well as with the patient group not diagnosed with dysarthria (Wilcoxon rank sum test, Ws = 394, z = –3.53, P , 0.001, r = –0.57). A significant difference in saccade duration was obtained between the GMFCS level 1 and level 4 groups (Wilcoxon rank sum test corrected for multiple comparisons, Ws = 216, z = –3.36, P , 0.001, r = –0.61). Even though the duration of fixations is, on average, not significantly different among the healthy, CP with no dysarthria, and CP with dysarthria groups (Figure 6), and significant differences occur for the duration of saccades (Figure 7), for both measures there is large variability, especially within each of the CP groups. Thus, it is important to analyze how

Figure 6. (A) Average duration of fixations (in seconds) for the different groups. (B) Average duration as a function of GMFCS level. (Symbols are described in Figure 5 legend.)

Abbreviation: GMFCS, Gross Motor Function Classification System.

these 2 measures relate at an individual level. This was done by analyzing the ratio of average fixation duration to average saccade duration within each individual participant. This ratio provides a measure, within each patient, of the relation between the time the patient requires for processing (individual average fixation duration) and the time required to search for new words (individual saccade duration). Larger ratio values indicate increased difficulties in processing the information read or shorter saccade durations. Results for the different groups as well as a function of GMFCS level are depicted in Figure 8. Although there was high variability across individual patients, in general lower ratios were obtained for CP patients, and particularly those


Lampe et al


Figure 7. (A) Average saccade duration (in seconds) for the different groups. (B) Average duration as a function of GMFCS level. (Symbols are described in Figure 5 legend.)


diagnosed with dysarthria (“ × ” in Figure 8A). Wilcoxon rank sum tests, corrected for the number of comparisons, showed significant differences between the healthy and dysarthria groups (Ws = 735, z = 4.74, P , 0.001, r = 0.75) as well as between the CP with dysarthria and CP with no dysarthria groups (Ws = 152, z = –2.56, P = 0.01, r = –0.42). No statistically significant differences were obtained between the healthy and no dysarthria groups. A 1-way ANOVA on the ratio of average fixation duration to average saccade duration for the CP group was used to test an effect of the GMFCS level. Wilcoxon rank sum tests corrected for the number of comparisons showed statistical significant differences between the GMFCS level 1 and level 4 groups (Ws = 361, z = –2.84, P = 0.004, r = –0.52). As shown on 154

Figure 8. (A) The ratio of average fixation duration to average saccade duration calculated for the 3 groups. (B) The ratio as a function of GMFCS level. (Symbols are described in Figure 5 legend.)


Figures 6 and 7, the lower individual ratios observed for CP patients in Figure 8 more likely reflect the longer saccade durations rather than shorter fixation times. Although not shown here, analysis of the average saccade duration and text comprehension index obtained for the only text that was read by all participants, text 1, showed no significant correlation between these 2 variables. Pupil size was also analyzed, as it can provide information about the cognitive effort to do the task.24–26 Figure 9 illustrates examples from 3 participants of how pupil size changed during text reading. Vertical black bars separate measures taken during the reading of different texts. Not all participants read all 5 texts. Also, only valid data, for which the eye-tracking data was uncorrupted, is shown. Gray and




Figure 9. (A–C) Pupil size during text reading for 3 participants.

black dots illustrate pupil size of the right and left eye, respectively. In Figure 9A, the pupil size decreased during the reading of single texts. As the text changed, the pupil size increased and a similar decrease in pupil size followed. In Figure 9B, the pupil size remained unchanged during reading. In Figure 9C, pupil size tended to increase during reading. Although not shown, all 3 types of changes in pupil size

were observed for the 3 groups, suggesting similar levels of attention across participants, independent of the group and of the severity of the CP condition (GMFCS level).

Discussion

This study used an eye-tracking system to record the movement of the eyes during a reading task in adult patients



Lampe et al

diagnosed with CP. In comparison with a group of healthy readers, CP patients took longer to read the texts, made more and shorter regressions, made shorter saccades, and showed lower ratios of average fixation duration to average saccade duration. Although the average duration of fixation did not differ significantly between the 2 groups, the duration of saccades was observed to be significantly longer for CP patients. No statistically significant effect of GMFCS level was observed on the reading rate or on the average duration of fixations, but it had a significant effect on the ratio of average fixation duration to average saccade duration and on the average saccade duration. Similar changes in pupil size during reading were observed for healthy and patients with CP. We found that the use of eyeglasses could affect the measurements, causing the eye-tracking system to lose the location of the pupil. As discussed by Blascheck,27 this may be caused by reflexive lenses; a hanging eyelid can also cause errors in eye tracking. Although 40% of the recorded data sets were considered faulty and not valid for further analysis, the eye-tracking method was nevertheless successful in capturing abnormal eye movements in the CP clinical population. By accurately following a person’s eye movements, eye-tracking systems provide a noninvasive way of assessing the neurocognitive processes that control this movement. It is not surprising, therefore, that eye-tracking systems have become an important tool in recent years in neuroscience and psychology research,28 and in particular in the study of reading abilities and the associated neurocognitive factors. Furthermore, eyetracking systems provide a communication channel that has been used in the development of human–computer interaction systems, particularly relevant when communication problems exist.29,30 Despite the numerous studies using this technique in the context of reading research and on particular in clinical populations,31 to our knowledge eye-tracking tools have not been used for research into reading skills in patients with CP. As described by Schneider and Kurt,32 difficulties experienced when learning to read can be demonstrated from the analysis of the direction of view, of regressions, and of the length of saccades while reading. Similar to their observations and in agreement with the previous literature on reading,19 histograms summarizing the horizontal spatial movement of the eyes during reading showed generally a higher number of saccades, a higher number of shorter saccades, as well as more regressions in the CP group than in the healthy group (Figure 4). The higher number of saccades and higher number of fixations suggest that CP readers need more often to bring text elements into foveal vision33,34 in order to understand the text, and hence it suggests that they 156

experience increased difficulty in processing the information. Skillful readers typically skip frequent function words (for a review, see Liversedge and Findlay35) and therefore in comparison present fewer fixations. Together with the higher number of short saccades, the higher number of fixations may also suggest a less efficient way of exploring the text field presented.36 In his work on language and writing problems in CP patients, Thiele37 observed that the anatomical brain damage in this group can lead to accommodation disturbances and problems of eyes fixation. In the current study, no significant increase in the average duration of fixation was observed in CP patients in comparison with healthy controls, and no significant effect of GMFCS level was observed. Nevertheless, the CP groups showed larger variability than the healthy group, with some CP patients presenting long fixation durations. Hence, given that the duration of fixation is indicative of the difficulty in extracting the information from the word being presented,19,38 and given that the average duration of fixation did not change significantly with GMFCS level (Figure 6B), the results then suggest that extracting the meaning of words was not particularly difficult for CP patients, in comparison with healthy controls. In this context, the fact that CP patients showed shorter regressions, which is generally interpreted as an indication that the reader has less meaningful information and therefore needs to reassess it, may be an indication of deficient short-term memory. The short-term memory allows readers to save the content fast and intentionally. The capacity for remembering words, letters, and numbers is restricted and relatively fixed.39 The lower fixation duration to saccade duration ratios observed for CP patients in this study were in the majority of cases due to longer saccade durations (Figures 6 to 8). The average duration of saccades did not correlate with text comprehension classification. The latter varied widely across CP patients, especially those with higher GMFCS levels. It is notable that patients with GMFCS level 5 also had short average saccades (Figure 7). Also remarkable is the fact that the patient whose results are presented in Figure 9A, suffering from hemiparesis, was able to read the 5 available texts correctly. The fact the average saccade duration did not correlate with text comprehension, together with the similarity in average fixation duration across GMFCS levels, suggests that the slower saccade movement is likely to reflect difficulties in searching visually through the text, rather than reflecting difficulties in processing the meaning of the words. This would be consistent with the reported difficulties by




children with CP in initiating and performing saccades and scanning the visual environment.7 Longer searching times in the CP group would also mean more memory demands, and therefore could also contribute to the need for making more regressions to remember information collected earlier. It is therefore possible that this clinical group may benefit from displays that facilitate the localization of the next target word during reading, making text reading a more efficient and probably a more enjoyable exercise. Hence, optimizing text design elements, such as type size and size of spaces between the words, can facilitate reading for this clinical population. Changes in pupil diameter have been associated with cognitive load in diverse tasks. Hess and Polt40 observed in multiplication tests that the diameter of the pupil increases with increasing difficulty of tasks. Pupillary response was also associated with processing load in reading tasks.26 Kahneman and Beatty41 proposed that the pupil size can be considered as a measure of mental capacity. Effects of stress42 and of emotional reactions43 have also been reported to affect pupillary response, besides the known effects of luminance. In the current study, pupillary diameter varied during the reading task; sometimes it increased from the beginning to the end of the text, sometimes it decreased, and sometimes it did not change. All of these variations were observed both for patients with CP and for healthy participants. Given the variety of factors that can affect the pupillary responses, it is difficult to draw definitive conclusions. Nevertheless, considering the above discussion, cognitive load seems to vary across healthy readers as well as across patients with CP. The appearance of speech disorders such as dysarthria correlates with the severity of disability. Cockerill et al44 reported on 346 young people aged 16 to 18 years with bilateral cerebral palsy; 63% had impaired speech of varying severity. Coleman et al45 found that preschool children with CP and with more severe gross motor impairment showed delayed communication, whereas children with mild motor impairment were less vulnerable. In our investigation, 65% of the severe disabled patients with GMFCS levels 4 and 5 had dysarthria, but they did not show increased difficulties in reading comprehension. Differences in the results between CP patients with and without dysarthria, therefore, can be attributed to the average higher GMFCS level in patients with dysarthria. The method of eye tracking has been described frequently in advertising, psychology, and control of devices.30 For patients with CP, eye-tracking methods have often been described in the context of the human–computer interaction

systems that facilitate communication abilities to support patients with communication and motor deficits.29,46 However, no previous studies have used eye tracking to investigate neurophysiological processes in patients with CP. In this study we found that eye tracking can also be used to investigate the reading abilities of CP patients. Knowledge of how patients with CP access and process information during reading would be particularly helpful in the development of methods that can facilitate their educational progress.

Conclusion

Objective recordings of eye movements during a reading task can be collected with an eye-tracking system in adults diagnosed with CP. The reading rate for patients with CP was lower than for healthy controls. A statistically significant effect of group, cerebral palsy versus healthy controls, and of GMFCS level was found. The ratio of average fixation duration to average saccade duration as well as the average duration of saccades was larger for CP patients in comparison with controls. The severity of motor impairment, classified according the GMFCS level, had a significant effect on both measures. The average durations of fixations did not differ significantly between patients with CP and the healthy participants nor with GMFCS level. Good text comprehension could also be achieved by CP patients with high GMFCS levels. The results suggest that CP patients may experience difficulties searching for words during reading. Eye-tracking systems yield better insight into visual perception and interpretation of texts in patients with early brain damage. The results reported here should be helpful for speech therapists and teachers in exercising and adapting text design.

Acknowledgments

Experimental parts of the study were financed by the Kraußianum Foundation.

Conflict of Interest Statement

Renée Lampe, MD, Varvara Turova, PhD, Tobias Blumenstein, MSc, and Ana Alves-Pinto, PhD, disclose no conflicts of interest.


Lampe et al


References 1. Bax M, Goldstein M, Rosenbaum P, et al. Proposed definition and classification of cerebral palsy. Dev Med Child Neurol. 2005;47(8):571–576. 2. Largo R. Frühkindliche Zerebralparese: Epidemiologische und klinische Aspekte [article in German]. Dtsch Arztebl. 1991;88(23):2061–2071. 3. Stotz S. Treatment of Cerebral Palsy. München: Pflaum Verlag; 2000. 4. Dufresne D, Dagenais L, Shevell MI. Spectrum of visual disorders in a population-based cerebral palsy cohort. Pediatr Neurol. 2014;50(4):324–328. 5. Fazzi E, Signorini SG, LA Piana R, et al. Neuro-ophthalmological disorders in cerebral palsy: ophthalmological, oculomotor, and visual aspects. Dev Med Child Neurol. 2012;54(8):730–736. 6. Guzzetta A, Mercuri E, Cioni G. Visual disorders in children with brain lesions: 2. Visual impairment associated with cerebral palsy. Eur J Paediatr Neurol. 2001;5(3):115–119. 7. Salati R, Borgatti R, Giammari G, Jacobson L. Oculomotor dysfunction in cerebral visual impairment following perinatal hypoxia. Dev Med Child Neurol. 2002;44(8):542–550. 8. Katayama M, Tamas LB. Saccadic eye-movement of children with cerebral palsy. Dev Med Child Neurol. 1987;29(1):36–39. 9. Foreman N, Fielder A, Minshell C, Hurrion E, Sergienko E. Visual search, perception, and visual-motor skill in “healthy” children born at 27–32 weeks’ gestation. J Exp Child Psychol. 1997;64(1):27–41. 10. Hood B, Atkinson J. Sensory visual loss and cognitive deficits in the selective attentional system of normal infants and neurologically impaired children. Dev Med Child Neurol. 2008;32(12):1067–1077. 11. Kozeis N, Anogeianaki A, Mitova DT, Anogianakis G. Visual function and execution of microsaccades related to reading skills, in cerebral palsied children. Int J Neurosci. 2006;116(11):1347–1358. 12. Kozeis N, Anogeianaki A, Mitova DT, Anogianakis G, Mitov T, Klisarova A. Visual function and visual perception in cerebral palsied children. Ophthalmic Physiol Opt. 2007;27(1):44–53. 13. Fedrizzi E, Anderloni A, Bono R, et al. Eye-movement disorders and visual-perceptual impairment in diplegic children born preterm: a clinical evaluation. Dev Med Child Neurol. 1998;40(10):682–688. 14. Saavedra S, Joshi A, Woollacott M, van Donkelaar P. Eye hand coordination in children with cerebral palsy. Exp Brain Res. 2009;192(2):155–165. 15. Richman JE, Garzia RP. Developmental Eye Movement Test, Examiner’s Booklet, Version 1. South Bend, IN: Bernell Corp; 1987. 16. Tassinari JT, DeLand P. Developmental eye movement test: reliability and symptomatology. Optometry. 2005;76(7):387–399. 17. Ayton LN, Abel LA, Fricke TR, McBrien NA. Developmental eye movement test: What is really measuring? Optom Vis Sci. 2009;86(6):722–730. 18. Starr MS, Rayner K. Eye movements during reading: some current controversies. Trends Cogn Sci. 2001;5(4):156–163. 19. Rayner K. Eye movements in reading and information processing: 20 years of research. Psychol Bull. 1998;124(3):372–422. 20. Lampe R, Mitternacht J, Gradinger R. Einsatz von Messinstrumenten zur Erkennung von Wahrnehmungsstörungen (Neuropaediatrie) [article in German]. Padiatr Padol. 2009;44(5):16–22. 21. Rosenbaum PL, Palisano RJ, Bartlett DJ, Galuppi BE, Russel DJ. Development of the gross motor function classification system for cerebral palsy. Dev Med Child Neurol. 2008;50(4):249–253. 22. Ghasia F, Brunstrom J, Gordon M, Tychsen L. Frequency and severity of visual sensory and motor deficits in children with cerebral palsy: gross motor function classification scale. Invest Ophthalmol Vis Sci. 2008;49(2):572–580. 23. Marques de Sá J.P. Applied Statistics: Using SPSS, STATISTICA, MATLAB and R. Berlin: Springer; 2007. 24. Beatty J, Wagoner BL. Pupillometric signs of brain activation vary with level of cognitive processing. Science. 1978;199(4334):1216–1218. 25. Hess EH. Attitude and pupil size. Sci Am. 1965;212(4):46–54. 26. Just MA, Carpenter PA. The intensity dimension of thought: pupillometric indices of sentence processing. Can J Exp Psychol. 1993;47(2):310–339.

158

27. Blascheck T. Eyetracking Basiertes Analysekonzept zur Evaluation von Visualisierungen. Diploma Thesis DIP−3302. Stuttgart: University of Stuttgart; 2012 [in German]. 28. Albert M, Heinz T, Imhof M. Entwicklung eines Mobilen Eyetrackers. University of Bamberg; November 2010 [in German]. http://www. cogsys.wiai.uni-bamberg.de /teaching/ss10/projekt/eyetracker.pdf. Accessed April 15, 2014. 29. Man DW, Wong MS. Evaluation of computer-access solutions for students with quadriplegic athetoid cerebral palsy. Am J Occup Ther. 2007;61(3):355–364. 30. Stoll J, Kohlbecher S, Marx S, Schneider E, Einhäuser W. Mobile three dimensional gaze tracking. In: Westwood JD, et al, eds, Studies in Health Technology and Informatics, Vol 163. Amsterdam: IOS Press; 2011:616–622. 31. Folta K, Mähler C. Schnelle Augenbewegungen und visuelle Fixation bei Kindern mit einer Aufmerksamkeitsdefizit-/Hyperaktivitätsstörung (ADHS) [article in German]. Kindheit Entwicklung. 2011;20(1):21–30. 32. Schneider G, Kurt J. Bücher sind gedruckte Arzneimittel. Lesen macht glücklich, da Lesen Mühe bereitet. http://www2.hu-berlin.de/reha/eye/ Blickbewegung%20beim%20 Lesen.pdf [in German]. Accessed April 14, 2014. 33. Rayner K, Bertera JH. Reading without a fovea. Science. 1979; 206(4417):468–469. 34. Rayner K, Inhoff AW, Morrison RE, Slowiaczek ML, Bertera JH. Masking of foveal and parafoveal vision during eye fixations in reading. J Exp Psychol Hum Percept Perform. 1981;7(1):167–179. 35. Liversedge SP, Findlay JM. Saccadic eye movements and cognition. Trends Cogn Sci. 2000;4(1):6–14. 36. Goldberg HJ, Kotval XP. Computer interface evaluation using eye movements: methods and constructs. Int J Ind Ergon. 1999;24:631–645. 37. Thiele A. Infantile Cerebralparese: zum Verhältnis von Bewegung, Sprache und Entwicklung—Theoretische Grundlagen einer Frühen Förderung Verbaler und Nonverbaler Kommunikation [in German]. Berlin: Edition Marhold; 1999. 38. Just MA, Carpenter PA. Eye fixations and cognitive processes. Cogn Psychol. 1976;8(4):441–480. 39. Miller GA. The magical number seven plus or minus two: some limits on our capacity for processing information. Psychol Rev. 1956;63(2):81–97. 40. Hess EH, Polt JH. Pupil size in relation to mental activity during simple problem solving. Science. 1964;143(3611):1190–1192. 41. Kahneman D, Beatty J. Pupil diameter and load on memory. Science. 1966;154(3756):1583–1585. 42. Zimmer K. Charakteristik von Informationsverarbeitungsaufwand und motivationaler Aktivierung über Kennwerte der Pupillenmotorik—ein Beitrag zur kognitiven Psychologie [dissertation in German]. Berlin: Humboldt Universität; 1984. 43. Johnson DA. Pupillary responses during a short-term memory task: Cognitive processing, or arousal, or both? J Exp Psychol. 1971;90(2):311–318. 44. Cockerill H, Elbourne D, Allen E, et al. Speech, communication and use of augmentative communication in young people with cerebral palsy: The SH&PE population study. Child Care Health Dev. 2013;40(2):149–157. 45. Coleman A, Weir KA, Ware RS, Boyd RN. Relationship between communication skills and gross motor function in preschool-aged children with cerebral palsy. Arch Phys Med Rehabil. 2013;94(11):2210–2217. 46. Amantis R, Corradi F, Molteni AM, et al. Eye-tracking assistive technology: is this effective for the developmental age? Evaluation of eyetracking systems for children and adolescents with cerebral palsy. In: Gelderblom GJ, Soede M, Adriaens L, Miesenberger K, eds. Everyday Technology for Independence and Care. Assistive Technologies Research Series, vol 29. Amsterdam: IOS Press; 2011:489–496.


Eye-Movement Control in RAN and Reading.

Eye movement alterations during reading in patients with early Alzheimer disease.

Human regional cerebral blood flow during rapid-eye-movement sleep.

Sedentary behaviour in adolescents and young adults with cerebral palsy.

Acquired cervical spine impairment in young adults with cerebral palsy.

Fast priming during eye fixations in reading.

The effects of crowding on eye movement patterns in reading.

Health Care Transition Experiences of Young Adults With Cerebral Palsy.

Focus on fatigue amongst young adults with spastic cerebral palsy.

Online Learners' Reading Ability Detection Based on Eye-Tracking Sensors.

Eye movements during reading: case reports.

Online Sentence Reading in People With Aphasia: Evidence From Eye Tracking.

Cerebral O2 metabolism and cerebral blood flow in humans during deep and rapid-eye-movement sleep.

Incrementality in Planning of Speech During Speaking and Reading Aloud: Evidence from Eye-Tracking.

Abnormal eye movement behaviour during text reading in neglect syndrome: a case study.

Number line estimation strategies in children with mathematical learning difficulties measured by eye tracking.

Factor Structure of Attention Capacities Measured With Eye-Tracking Tasks in 18-Month-Old Toddlers.

Vermal infarct with pursuit eye movement disorders.

Eye Movement Patterns in Natural Reading: A Comparison of Monolingual and Bilingual Reading of a Novel.

Form-to-expectation matching effects on first-pass eye movement measures during reading.

Reading during the composition of multi-sentence texts: an eye-movement study.

Processing Rhythmic Pattern during Chinese Sentence Reading: An Eye Movement Study.

Return sweep latency in reading with a paralyzed eye.

Assessing cerebral glucose metabolism in patients with idiopathic rapid eye movement sleep behavior disorder.