BRIEF REPORTS MILITARY MEDICINE, 179, 6;6I9. 2014
Predictive Visual Tracking: Specificity in Mild Traumatic Brain Injury and Sleep Deprivation Jun Maruta, PhD"; Kristin J. Heaton, PhDff, Alexis L Maule, MPHft; Jamshid Ghajar, MD, PhD*§ ABSTRACT We tested whether reduced cognitive function associated with mild traumatic braiti injury (mTBI) and sleep deprivation can be detected and distinguished using indices of predictive visual tracking. A circular visual tracking test was giveti to 13 patients with acute mTBI (recruited within 2 weeks of injury), 127 normal control subjects, and 43 healthy subjects who were fatigued by 26-hour sleep deprivation. Eye movement was monitored with videooculography. In the mTBI-related portion of the study, visual tracking performance of acute mTBI patients was significantly worse than normal subjects (p < 0.001). In the sleep-deprivation-related portion of the study, no chatige was detected between the two baseline measures separated by 2 to 3 weeks, but the 26-hour sleep deprivation significantly degraded the visual tracking performance {p < 0.001). The mTBI subjects had substantially worse visual tracking than sleep-deprived subjects that could also be identified with different visual tracking indices, indicating possible different neurophysiological mechanisms. Results suggest that cognitive impairment associated with mTBI and fatigue may be triaged with the aid of visual tracking measures.
INTRODUCTION The ability to focus attention may be compromised by various causes. Two of the mo.st commonly recognized conditions associated with reduced attention are mild traumatic brain injury (mTBI)'-^ and fatigue induced by sleep deprivaBoth of these conditions are pertinent to military operations because of their prevalence and effects on soldier cognitive performance. Correct identification of degraded attention from either mTBI or fatigue is necessary in triaging cognitive impairment. In this study, we tested whether the alteration in attention associated with mTBI and sleep deprivation can be detected with indices of predictive visual tracking performance. This supposition is based on the grounds that eye movement and attention processes are implemented by closely overlapping areas of the brain,'^ and visual tracking perfoiTnance is dependent upon attention.^'^ Furthermore, because of the visuomotor processing delay, successful visual tracking requires dynamic *Brain Trauma Foundation, 7 World Trade Center, 34th Floor, 250 Greenwich Street, New York. NY 10007. tMilitary Performance Division, United States Army Research Institute of Environmental Medicine, Kansas Street, Building 42, Natick, MA 01760. iDepartment of Environmental Health. Boston University School of Public Health. 715 Albany Street, Talbot Building, Boston, MA 02118. §Department of Neurosurgery, Stanford University School of Medicine, 300 Pa.steur Drive, Stanford, CA 94305. The opinions or assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the views of the Army or the Department of Defense. doi; 10.7205/MILMED-D-13-00420
MILITARY MEDICINE, Vol. 179, June 2014
synchronization of the internally generated prediction with the external stimulus. This dynamic visuomotor synchronization is supported by the predictive timing function, which is considered to be an essential element of attention.** Since attention performance can be diminished by mTBI or sleep deprivation, we compared the performance of predictive visual tracking in subject groups with these conditions and respective controls using multiple indices derived for the circular visual tracking protocol that we have developed.'' The continuous circular motion of the stimulus is well suited for characterizing maintenance of visual tracking and likely sensitive to attention impairment. MATERIALS AND METHODS The present prospective study was conducted at the Citigroup Biomédical Imaging Center at Weill Cornell Medical College (WCMC) in New York, New York and at the United States Army Research Institute of Environmental Medicine (USARIEM) located at the Natick Soldier Center, Natick, Massachusetts. The protocol was reviewed and approved by the WCMC Institutional Review Board, the USARIEM Human Use Review Committee, and the USARIEM Office of Research Ouality and Compliance. Written informed consent was obtained from all participants before data collection. Subjects mTBI Suhjects and Normal Controis
A total of 13 civilian subjects with a history of mTBI (a loss of consciousness less than 30 minutes and post-traumatic
619
Predictive Visual Tracking in mTBI and Sleep Deprivation
amnesia less than 24 hours) and active postconcussive symptoms were recruited through referrals from physicians within 2 weeks of injury. The age range was from 13 to 41 years old (mean 19.6) and there were eight males and five females. To provide a basis for normative comparisons, healthy civilian control subjects with no prior history of head injury, who had completed at least 12 years of education, were also recruited. None of the control or mTBI subjects were pregnant; had a history of drug or alcohol abuse, neurological or psychiatric illness, or seizure; or had gross visual (worse than 20/30 corrected or uncorrected) or hearing problems. Of the 127 normal subjects recruited (aged 18-55 years), 57 were male (mean age 34.3) and 70 were female (mean age 36.6). We did not have a pédiatrie control group. Thus, although functional maturity of the smooth pursuit eye movement system is thought to be achieved by mid-adolescence,"'" the normative comparison for eight subjects aged 13 to 17 may be slightly skewed. The patients and normal control subjects were tested at WCMC. An outline of these patients' recovery process up to 1-month postinjury has been published previously. 13 Sleep-Deprivation Subjects
A total of 43 healthy military volunteer subjects participated in the sleep-deprivation experiment at USARIEM, of whom 11 were female (mean age 22.2) and 32 were male (mean age 20.7). Potential participants were recruited at USARIEM via scheduled, in-person briefings. Participation was limited to men and women 18 to 50 years of age, who were able to abstain from caffeine use for at least 26 hours. Other eligibility criteria were the same as the civilian controls for the mTBI study. Four measurements were taken from this group of subjects. A first baseline measurement was taken 2 to 3 weeks before (Time 1) the sleep-deprivation period, and a second baseline measurement was taken (Time 2) during the initial period of sustained wakefulness. The two baseline tests and the last measurement (Time 4) took place in the morning (6;30 a.m.-9:3O a.m.) to control for circadian effects and to coincide with subjects' typical morning schedules. The third measurement (Time 3) took place at predawn (2;00 a.m.4;00 a.m.). The results of normal baseline and test-retest analyses of these subjects' visual tracking performance are described in another publication.'^
Eye Movement Recording and Scoring The eye movement recording and analysis procedures are detailed in another publication.'^ Briefly, the visual tracking testing protocol was implemented on an integrated stimulus presentation-eye tracking apparatus (EyeLink CL; SR Research, Ontario, Canada) with which eye movements were recorded at 500 Hz using infrared video-oculography while the head was stabilized with a head- and chinrest. The testing protocol was automated with each step in the experiment proceeded
620
with preprogrammed audio and visual instructions, whereas interventions by the experimenter were reduced to entering the data file name, optimizing video images of the eyes, initiating the calibration procedure, and ensuring valid calibration. The experimenters at WCMC and USARIEM were cross-trained. All subjects had normal or corrected-to-normal vision. The test stimulus was a red 0.5° target that moved clockwise in a circular trajectory of 10° radius in visual angle at 0.4 Hz against a black background on a computer screen. The stimulus amplitude and frequency fell in the range within which progressive degradation of performance occurs in normal subjects.''* Two consecutive runs of 6 cycles of circular movement were presented, each run lasting 15 seconds. The target and eye data were stored for offline analyses. The stability of the gaze on the target was characterized by the variability of the instantaneous gaze positional error, measured in visual angle, in the directions orthogonal and parallel to the target movement (standard deviation of radial errors [SDRE] and standard deviation of tangential errors [SDTE]). The larger the SDRE or SDTE value, the less stable the tracking. Overall spatial accuracy was measured with mean radius, and overall temporal accuracy was measured with mean phase error, with positive phase error defined as the gaze leading the target. We also report horizontal and vertical smooth pursuit velocity gain (H and V gains), which represent the ratios between the smooth pursuit eye velocity and the target velocity in the horizontal and vertical directions. A smaller gain indicates less precise tracking. Eye velocity was obtained by two-point differentiating eye-position data, and saccades were replaced with straight lines connecting the remaining segments before smooth pursuit velocity was calculated through a sine curve fit. The data from the first stimulus cycle were not analyzed because the segment contained the initial transient response to the target movement. Scoring was automated and conducted without any consideration of other clinical information.
Statisticai Anaiysis Box plots were used to visually compare distributions of visual tracking parameters. The bottom and top of each box indicate the first (Ql ) and third (Q3) quartiles, and the middle line the median. The whiskers indicate the minimum and maximum of all the data excluding outliers whose values were larger than Q^ + w (Qj, - Q\) or smaller than Qi M' (0.-Î - öt). where Q^ and Q^ are the 25th and 75th percentiles, respectively, and M' = 1.5. Two-sample i-tests and the area under the curve (AUC) of receiver-operator characteristic ' '' were used to detect differences in the visual tracking indices from mTBI patients and normal subjects. The a level was set aX p = 0.05. To test whether the inclusion of the eight adolescent mTBI subjects introduced a significant bias in the results that could be explained as normal age-related differences, their
MILITARY MEDICINE, Vol. 179, June 2014
Predictive Visual Tracking in niTBl and Sleep Deprivation
visual tracking performance was separately analyzed using linear regression. To identify the changes in performance between the sleepand nonsleep-deprived states, a repeated measures analysis of variance (ANOVA) with testing time as the factor (degrees of freedom [df] = 3, 42) and a post hoc test (Holm) were used. The receiver-operator characteristic AUC was also used to identify the changes in performance between the sleep- and nonsleep-deprived states. To examine the characteristics of spatial and temporal variability of visual tracking performance in different groups, a linear regression equation was derived for the relationship between SDRE and SDTE. Group differences in the slopes of the regression lines were tested with a f-test. Because there were two baseline measurements in the sleep-deprivation portion of the study, within-individual average baseline scores were obtained before the comparison. RESULTS
Mild Traumatic Brain Injury Group comparisons were made between normal control subjects and mTBI patients at the acute stage. An example of visual tracking performance of an mTBI patient is shown in Figure 1. As typical of degraded tracking in mTBI patients, the phase error was modulated with a sawtooth waveform with saccades that placed the gaze ahead of the target (phase error > 0) while spatial prediction was retained. Two-sample r-tests revealed that the acute mTBI group was significantly different from the normal group in all six visual tracking
mTBI 10 5 0 -5 -10 -10
SDRE = 0.93° SDTE = 2.31° Mean Radius = 9.99° Mean Phase = 4.90° 1s
20' -20'
k
FIGURE 1. Circular visual tracking after concussion. (Top left) Twodimensional gaze trajectory. (Top right) Gaze position in target-based reference frame, with the target fixed at the 12-o'clock position. The center of the white circle indicates the average gaze position. The dot-dashed curve indicates the circular path. Note the large asymmetry between the error variability in the directions parallel and orthogonal to the target trajectory. (Bottom) Time plot of phase error relative to the target. The discontinuations of the trace indicate blinks. The subject was a 21-year-old male student who 10 days before testing collided with another player during a soccer game and had post-traumatic amnesia.
MILITARY MEDICINE, Vol. 179, June 2014
indices (Table I). Specifically, mTBI patients exhibited increased gaze position error variability, reduced radius of gaze trajectory, phase leading, and reduced smooth pursuit velocity relative to healthy controls (Fig. 2). However, although statistically significant, the decrease in mean radius was only by 2% and did not explain the larger decrease in velocity gains. Within the eight adolescent mTBI subjects aged between 13 and 17, none of the visual tracking indices showed a significant linear dependence on age (0.10
Predictive Visual Tracking: Specificity in Mild Traumatic Brain Injury and Sleep Deprivation Jun Maruta, PhD"; Kristin J. Heaton, PhDff, Alexis L Maule, MPHft; Jamshid Ghajar, MD, PhD*§ ABSTRACT We tested whether reduced cognitive function associated with mild traumatic braiti injury (mTBI) and sleep deprivation can be detected and distinguished using indices of predictive visual tracking. A circular visual tracking test was giveti to 13 patients with acute mTBI (recruited within 2 weeks of injury), 127 normal control subjects, and 43 healthy subjects who were fatigued by 26-hour sleep deprivation. Eye movement was monitored with videooculography. In the mTBI-related portion of the study, visual tracking performance of acute mTBI patients was significantly worse than normal subjects (p < 0.001). In the sleep-deprivation-related portion of the study, no chatige was detected between the two baseline measures separated by 2 to 3 weeks, but the 26-hour sleep deprivation significantly degraded the visual tracking performance {p < 0.001). The mTBI subjects had substantially worse visual tracking than sleep-deprived subjects that could also be identified with different visual tracking indices, indicating possible different neurophysiological mechanisms. Results suggest that cognitive impairment associated with mTBI and fatigue may be triaged with the aid of visual tracking measures.
INTRODUCTION The ability to focus attention may be compromised by various causes. Two of the mo.st commonly recognized conditions associated with reduced attention are mild traumatic brain injury (mTBI)'-^ and fatigue induced by sleep deprivaBoth of these conditions are pertinent to military operations because of their prevalence and effects on soldier cognitive performance. Correct identification of degraded attention from either mTBI or fatigue is necessary in triaging cognitive impairment. In this study, we tested whether the alteration in attention associated with mTBI and sleep deprivation can be detected with indices of predictive visual tracking performance. This supposition is based on the grounds that eye movement and attention processes are implemented by closely overlapping areas of the brain,'^ and visual tracking perfoiTnance is dependent upon attention.^'^ Furthermore, because of the visuomotor processing delay, successful visual tracking requires dynamic *Brain Trauma Foundation, 7 World Trade Center, 34th Floor, 250 Greenwich Street, New York. NY 10007. tMilitary Performance Division, United States Army Research Institute of Environmental Medicine, Kansas Street, Building 42, Natick, MA 01760. iDepartment of Environmental Health. Boston University School of Public Health. 715 Albany Street, Talbot Building, Boston, MA 02118. §Department of Neurosurgery, Stanford University School of Medicine, 300 Pa.steur Drive, Stanford, CA 94305. The opinions or assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the views of the Army or the Department of Defense. doi; 10.7205/MILMED-D-13-00420
MILITARY MEDICINE, Vol. 179, June 2014
synchronization of the internally generated prediction with the external stimulus. This dynamic visuomotor synchronization is supported by the predictive timing function, which is considered to be an essential element of attention.** Since attention performance can be diminished by mTBI or sleep deprivation, we compared the performance of predictive visual tracking in subject groups with these conditions and respective controls using multiple indices derived for the circular visual tracking protocol that we have developed.'' The continuous circular motion of the stimulus is well suited for characterizing maintenance of visual tracking and likely sensitive to attention impairment. MATERIALS AND METHODS The present prospective study was conducted at the Citigroup Biomédical Imaging Center at Weill Cornell Medical College (WCMC) in New York, New York and at the United States Army Research Institute of Environmental Medicine (USARIEM) located at the Natick Soldier Center, Natick, Massachusetts. The protocol was reviewed and approved by the WCMC Institutional Review Board, the USARIEM Human Use Review Committee, and the USARIEM Office of Research Ouality and Compliance. Written informed consent was obtained from all participants before data collection. Subjects mTBI Suhjects and Normal Controis
A total of 13 civilian subjects with a history of mTBI (a loss of consciousness less than 30 minutes and post-traumatic
619
Predictive Visual Tracking in mTBI and Sleep Deprivation
amnesia less than 24 hours) and active postconcussive symptoms were recruited through referrals from physicians within 2 weeks of injury. The age range was from 13 to 41 years old (mean 19.6) and there were eight males and five females. To provide a basis for normative comparisons, healthy civilian control subjects with no prior history of head injury, who had completed at least 12 years of education, were also recruited. None of the control or mTBI subjects were pregnant; had a history of drug or alcohol abuse, neurological or psychiatric illness, or seizure; or had gross visual (worse than 20/30 corrected or uncorrected) or hearing problems. Of the 127 normal subjects recruited (aged 18-55 years), 57 were male (mean age 34.3) and 70 were female (mean age 36.6). We did not have a pédiatrie control group. Thus, although functional maturity of the smooth pursuit eye movement system is thought to be achieved by mid-adolescence,"'" the normative comparison for eight subjects aged 13 to 17 may be slightly skewed. The patients and normal control subjects were tested at WCMC. An outline of these patients' recovery process up to 1-month postinjury has been published previously. 13 Sleep-Deprivation Subjects
A total of 43 healthy military volunteer subjects participated in the sleep-deprivation experiment at USARIEM, of whom 11 were female (mean age 22.2) and 32 were male (mean age 20.7). Potential participants were recruited at USARIEM via scheduled, in-person briefings. Participation was limited to men and women 18 to 50 years of age, who were able to abstain from caffeine use for at least 26 hours. Other eligibility criteria were the same as the civilian controls for the mTBI study. Four measurements were taken from this group of subjects. A first baseline measurement was taken 2 to 3 weeks before (Time 1) the sleep-deprivation period, and a second baseline measurement was taken (Time 2) during the initial period of sustained wakefulness. The two baseline tests and the last measurement (Time 4) took place in the morning (6;30 a.m.-9:3O a.m.) to control for circadian effects and to coincide with subjects' typical morning schedules. The third measurement (Time 3) took place at predawn (2;00 a.m.4;00 a.m.). The results of normal baseline and test-retest analyses of these subjects' visual tracking performance are described in another publication.'^
Eye Movement Recording and Scoring The eye movement recording and analysis procedures are detailed in another publication.'^ Briefly, the visual tracking testing protocol was implemented on an integrated stimulus presentation-eye tracking apparatus (EyeLink CL; SR Research, Ontario, Canada) with which eye movements were recorded at 500 Hz using infrared video-oculography while the head was stabilized with a head- and chinrest. The testing protocol was automated with each step in the experiment proceeded
620
with preprogrammed audio and visual instructions, whereas interventions by the experimenter were reduced to entering the data file name, optimizing video images of the eyes, initiating the calibration procedure, and ensuring valid calibration. The experimenters at WCMC and USARIEM were cross-trained. All subjects had normal or corrected-to-normal vision. The test stimulus was a red 0.5° target that moved clockwise in a circular trajectory of 10° radius in visual angle at 0.4 Hz against a black background on a computer screen. The stimulus amplitude and frequency fell in the range within which progressive degradation of performance occurs in normal subjects.''* Two consecutive runs of 6 cycles of circular movement were presented, each run lasting 15 seconds. The target and eye data were stored for offline analyses. The stability of the gaze on the target was characterized by the variability of the instantaneous gaze positional error, measured in visual angle, in the directions orthogonal and parallel to the target movement (standard deviation of radial errors [SDRE] and standard deviation of tangential errors [SDTE]). The larger the SDRE or SDTE value, the less stable the tracking. Overall spatial accuracy was measured with mean radius, and overall temporal accuracy was measured with mean phase error, with positive phase error defined as the gaze leading the target. We also report horizontal and vertical smooth pursuit velocity gain (H and V gains), which represent the ratios between the smooth pursuit eye velocity and the target velocity in the horizontal and vertical directions. A smaller gain indicates less precise tracking. Eye velocity was obtained by two-point differentiating eye-position data, and saccades were replaced with straight lines connecting the remaining segments before smooth pursuit velocity was calculated through a sine curve fit. The data from the first stimulus cycle were not analyzed because the segment contained the initial transient response to the target movement. Scoring was automated and conducted without any consideration of other clinical information.
Statisticai Anaiysis Box plots were used to visually compare distributions of visual tracking parameters. The bottom and top of each box indicate the first (Ql ) and third (Q3) quartiles, and the middle line the median. The whiskers indicate the minimum and maximum of all the data excluding outliers whose values were larger than Q^ + w (Qj, - Q\) or smaller than Qi M' (0.-Î - öt). where Q^ and Q^ are the 25th and 75th percentiles, respectively, and M' = 1.5. Two-sample i-tests and the area under the curve (AUC) of receiver-operator characteristic ' '' were used to detect differences in the visual tracking indices from mTBI patients and normal subjects. The a level was set aX p = 0.05. To test whether the inclusion of the eight adolescent mTBI subjects introduced a significant bias in the results that could be explained as normal age-related differences, their
MILITARY MEDICINE, Vol. 179, June 2014
Predictive Visual Tracking in niTBl and Sleep Deprivation
visual tracking performance was separately analyzed using linear regression. To identify the changes in performance between the sleepand nonsleep-deprived states, a repeated measures analysis of variance (ANOVA) with testing time as the factor (degrees of freedom [df] = 3, 42) and a post hoc test (Holm) were used. The receiver-operator characteristic AUC was also used to identify the changes in performance between the sleep- and nonsleep-deprived states. To examine the characteristics of spatial and temporal variability of visual tracking performance in different groups, a linear regression equation was derived for the relationship between SDRE and SDTE. Group differences in the slopes of the regression lines were tested with a f-test. Because there were two baseline measurements in the sleep-deprivation portion of the study, within-individual average baseline scores were obtained before the comparison. RESULTS
Mild Traumatic Brain Injury Group comparisons were made between normal control subjects and mTBI patients at the acute stage. An example of visual tracking performance of an mTBI patient is shown in Figure 1. As typical of degraded tracking in mTBI patients, the phase error was modulated with a sawtooth waveform with saccades that placed the gaze ahead of the target (phase error > 0) while spatial prediction was retained. Two-sample r-tests revealed that the acute mTBI group was significantly different from the normal group in all six visual tracking
mTBI 10 5 0 -5 -10 -10
SDRE = 0.93° SDTE = 2.31° Mean Radius = 9.99° Mean Phase = 4.90° 1s
20' -20'
k
FIGURE 1. Circular visual tracking after concussion. (Top left) Twodimensional gaze trajectory. (Top right) Gaze position in target-based reference frame, with the target fixed at the 12-o'clock position. The center of the white circle indicates the average gaze position. The dot-dashed curve indicates the circular path. Note the large asymmetry between the error variability in the directions parallel and orthogonal to the target trajectory. (Bottom) Time plot of phase error relative to the target. The discontinuations of the trace indicate blinks. The subject was a 21-year-old male student who 10 days before testing collided with another player during a soccer game and had post-traumatic amnesia.
MILITARY MEDICINE, Vol. 179, June 2014
indices (Table I). Specifically, mTBI patients exhibited increased gaze position error variability, reduced radius of gaze trajectory, phase leading, and reduced smooth pursuit velocity relative to healthy controls (Fig. 2). However, although statistically significant, the decrease in mean radius was only by 2% and did not explain the larger decrease in velocity gains. Within the eight adolescent mTBI subjects aged between 13 and 17, none of the visual tracking indices showed a significant linear dependence on age (0.10
