Longitudinal Study on the Detection and Evaluation of Onset Mild

Traumatic Brain Injury during Dual Motor and Cognitive Tasks

Hung Nguyen

, Fouaz Ayachi

, Etienne Goubault

, Catherine Lavigne-Pelletier

Bradford J. McFadyen

2,3

and Christian Duval

1,4

Département des Sciences des I’activité physique, Université du Québec à Montréal, Montréal, Canada

Département de Réadaptation, Faculté de Médecine, Université laval, Québec, Canada

Centre interdisciplinaire de recherche en réadaptation et intégration sociale, Québec, Canada

Centre de Recherche Institut Universitaire de Gériatrie de Montréal, Montreal, Canada

Keywords: Concussion, Youth, Athletes, Markerless, Football, Walking, Obstacle, Performance, Speed, Step Width,

Preventive, Sensors, Head Injuries.

Abstract: Currently, concussions are detected by observing physical and cognitive symptoms such as dizziness,

disorientation and loss of consciousness that are often associated with mild traumatic brain injury (mTBI).

Evaluation methods such as neurocognitive tests and neuroimaging are often performed post-concussion.

However, these methods can be expensive and cumbersome to use. In this study, we developed a new

testing protocol using a markerless motion capture system to quickly monitor the cognitive and motor

dysfunction of football players over the course of the season. This protocol utilized a dual-task paradigm to

identify kinematic measures that could detect the subtle changes in the motor and cognitive function of

players due to mTBI. Four high school football players (2 healthy and 2 with history of concussion)

volunteered to participate in the study. Participants were asked to navigate a staged obstacle course with and

without an N-Back (N-2) cognitive task. Positional data of 23 limb segment nodes were recorded using

markerless motion tracking system. Data collection lasted less than 5 minutes, with minimal preparation

time. The results showed that walking speed, median frequency of sacrum in the vertical direction and step

width variability during straightway walking were strongly associated with the presence of mTBI.

1 INTRODUCTION

Among high school athletes in the U.S, football

players have the highest rate of concussion at 15%

per season (Rosenthal et al., 2014). However, the

true incidence is estimated to be much higher due to

the difficulty of detecting concussion as well as

under reporting. Recently, concussion in sports has

garnered tremendous media attention due to the

catastrophic effects on the health of retired

professional football players (Pilon and Belson,

2013) as they deal with consequences of repetitive

impact on their body throughout their football

career. While the attention to concussion has

encouraged team and medical professionals to be

more cautious when administering post-concussion

protocols, there are still elements of subjective

diagnoses involving concussion. Furthermore, the

reactive diagnosis of concussion ignores the

progressive aging of the brain due to repetitive

impacts (Bowen, 2003, Bey and Ostick, 2009, Cantu

and Gean, 2010). These repeated impacts often

induced more long-term problems due to our

inability to quantify the effect of concussion. In

youth athletics, where resources are often

unavailable to monitor the impact of these hits on

their developing brain (Field et al., 2003, Grady,

2010), a quick and accurate testing mechanism is

needed to arm healthcare professionals with

quantifiable metrics to evaluate the cognitive and

motor dysfunction of players who are at high risk of

concussion. If we can longitudinally monitor the

players and detect subtle changes in their behaviour

that correlate to traumatic brain injury, it is possible

to prescribe preventive care that could minimize the

long-term effects of mild traumatic brain injury

(mTBI).

Currently, concussion is diagnosed by visually

Nguyen, H., Ayachi, F., Goubault, E., Lavigne-Pelletier, C., McFadyen, B. and Duval, C..

Longitudinal Study on the Detection and Evaluation of Onset Mild Traumatic Brain Injury during Dual Motor and Cognitive Tasks.

In Proceedings of the 3rd International Congress on Sport Sciences Research and Technology Support (icSPORTS 2015), pages 77-83

ISBN: 978-989-758-159-5

observing signs that often associate with mTBI such

as headache, disorientation, dizziness and more

severely, loss of consciousness. However, loss of

consciousness only happens in less than 5% of

reported concussion (Castile et al., 2012). These

symptoms make concussion difficult to objectively

quantify. The diagnostics and follow-up of clinical

concussion are normally done by administering

neurocognitive and balance tests such as the Sports

Concussion Assessment Tool (SCAT3) that

subjectively scores the cognitive and motor

performance of an individual based on

questionnaires and the results of simple motor and

cognitive tasks. More advance methods of

diagnosing concussion such as Functional Magnetic

Resonance Imaging (fMRI) has also been used to

evaluate cranial fracture, bleeding (Khurana and

Kaye, 2012) and monitor of the alteration in the

brain (Talavage et al., 2014). However, fMRI are

expensive, time consuming and not widely available.

Accordingly, it would be impossible to administer

these tests chronically to monitor the state of the

player during the season. The diagnostic value of

such tests could also be challenged since most

concussions do not show enough structural damage

to be detected by neuroimaging.

Recently, more efforts are focused on embedding

sensors, such as accelerometers, into helmets to

longitudinally monitor the cranial movement during

exposure to impacts. The aim is to use these sensors

to develop performance metrics to determine if an

impact experienced by a player could lead to clinical

concussion. However, the issue is complex and it

had been shown that acceleration alone could not

predict concussion (Greenwald et al., 2008, Post and

Hoshizaki, 2015). For example, Talavage (2014)

used a combination of hit events (6 accelerometers

in the helmet), neurocognitive test and neuroimaging

to monitor the neurological and mental health of

varsity and junior varsity football players from pre

to post-season. The data showed that there was no

correlation between the peak helmet acceleration

and cognitive deficit. Auerbach (2015) used multiple

cranial accelerometers attached to the player’s

helmet to measure the pulsation of the cerebral

blood flow and its impact at different locations of

the skull. The results showed that there was a shift in

the high harmonic frequency of these sensors in the

concussed group. However, using a system of

sensors only monitor the physical impact to the head

of the individual while the cognitive performance is

ignored during process. Furthermore, the cause of

this shift in harmonic frequency remains

unexplained.

In this study, we proposed a quick and non-

invasive testing protocol using a dual-task paradigm

to extract relevant kinematics measurements that

could be used to longitudinally monitor the cognitive

and motor dysfunction of players who are at high

risk of concussion. Dual tasks that involved both

cognitive and motor tasks have demonstrated

promising results in differentiating the kinematic

behaviour between concussed and non-concussed

group. Cossette (2014) showed that gait speed and

dual-task cost during stepping over obstacle were

sensitive to the executive dysfunction of people with

mTBI. Parker (2005) showed gait and center of mass

measurement were significantly different between

the two groups during walking with a cognitive task.

A systematic review of the dual-task paradigm in

detecting concussion also concluded that gait

velocity and the sway in the medial lateral direction

could be used for long-term monitoring of sport-

related concussion (Lee et al., 2013). These results

in dual-task paradigm provide a viable platform to

develop a novel testing protocol that could quickly

and reliably monitor the cognitive and motor

behaviour of these high-risk players. Interestingly,

dual-tasking has a clear advantage compared to other

diagnostic methods; it replicates situations

encountered during play. Indeed, players are

systematically required to move their body onto the

playing field while making quick decisions. Their

ability to maintain their concentration, which is

normally associated with high cognitive load, is vital

in order to play the game. In fact, a reduction in

dual-tasking ability could put the player at risk of

further unnecessary impacts due to loss of situational

awareness on the field (Fait et al., 2013).

Contrary to previous studies where optical

systems requiring complicated setup and lengthy

testing protocol, the present study used a markerless

motion tracking system that is simple to setup. The

protocol was designed to minimize the disruption to

the player’s life while monitoring their cognitive and

motor performance throughout the season. Such a

system may eventually provide a platform to follow

players during an entire season, so as to help athletic

trainers and healthcare professionals detect and

manage concussions.

2 METHODS

2.1 Participants

Four male football players volunteered from a local

high school for the study (15.5 ± 0.6 years old,

icSPORTS 2015 - International Congress on Sport Sciences Research and Technology Support

height =1.82 ± 0.07 m, weight = 79.3 ± 13.0 kg).

Participants were excluded if they suffered physical

injury during the season that compromised their

normal gait pattern. Two participants had a history

of concussion while two had never been diagnosed

with concussion. During the course of the season,

one participant was clinically diagnosed with a

concussion. The institutional research ethics review

board of the Centre de Recherche de l’Institut

Universitaire de Gériatrie de Montreal (CRIUGM)

approved this research and each participant and their

guardians read and signed an informed consent

form.

2.2 Data Acquisition

A markerless motion capturing system from Organic

Motion (New York, NY) was used to monitor and

measure the performance of the football players

throughout the playing season. The system contains

16 cameras to capture the motion of the participants

within the staged course. The system generates a full

body skeletal model by tracking 23 nodes on the

body. The (x,y,z) coordinates of these nodes are

tracked by the system at 60 Hz. The data were

filtered using a zero-phase fourth-order digital filter

with a 6 Hz cut-off frequency. An external data

acquisition system controlled by DasyLab (Norton,

MA, USA) was also programed to randomly

generate numbers between 1 and 9 for the cognitive

tests (see below). The acquisition system also

monitored sensors attached to the turn markers and

the hanging obstacle on the testing stage to monitor

errors during turning (contact). The sensors data

were sampled at 100 Hz. All data were exported to

Matlab® (Natick, MA, USA) for analysis.

2.3 Protocol

Participants were asked to navigate a staged obstacle

course at higher than normal walking speed (without

running) for 45 seconds (Figure 1). The obstacle

course included tasks such as straightway walking,

turning and stepping under and over obstacles.

The testing stage had two obstacles: a ground

obstacle and a hanging obstacle. These obstacles

allowed us to measure the motor function in

stepping over and ducking under an obstacle. The

testing stage also had three turns. At each turn, there

was a strain gauge sensor attached to a string

marking the centroid of the turn. These sensors

recorded if the participants touched the string while

turning. Similarly, the hanging obstacle was also

equipped with a strain gauge to measure the

disturbance if participants touched the bar.

Figure 1: A) Top view schematic of the different

challenging segments in the testing protocol. B) 3-D view

of the testing stage.

An N-back (N-2) cognitive task (Kirchner, 1958)

was used to challenge the cognitive and motor

functions of the players. The test aims at evaluating

the working memory, concentration and their impact

on motor functions during the single and dual-task

environments. The dual-task was optimized to

challenge the players during the navigation of the

obstacle course. Such dual-task is ideal since it

replicates game situations where players are asked to

move on the field (motor aspect) while make

decisions according to their environment (cognitive

aspect). Participants were monitored throughout the

football season and tested at least once a week.

Participants were asked to perform two trials of the

cognitive task only, two trials of the motor task only,

one in each direction (clockwise and counter

clockwise) and, similarly, two trials of the motor and

cognitive tasks combined. It is important to note that

each testing session lasted on average less than 15

minutes from the time the participant entered the

laboratory and the time the test was completed. The

actual data collection was only 4.5 minutes.

2.4 Performance Analysis

2.4.1 Task Segmentation

The obstacle course was separated into different

sections for analysis. The segmentation was done by

specifying the x-y boundary on the stage to mark

different regions. Each segment provided specific

metrics for performance analysis corresponding to

the activities within each segment. For instance,

during the hanging obstacle segment, the distance

between the participant and the obstacle (clearance)

was used as a possible marker sensitive to change

over the course of the season. Similarly walking

speed and step width were only calculated during

straightway walking segment.

Longitudinal Study on the Detection and Evaluation of Onset Mild Traumatic Brain Injury during Dual Motor and Cognitive Tasks

2.4.2 Kinematics Analysis

While many kinematic performance parameters

could be used to determine whether they were viable

factors in discriminating the symptoms of mild

traumatic brain injury, this paper focused on

segment task time (of each segment was defined by

tracking when the sacrum entered and exited the x-y

boundary of each segment), median frequency (the

spectral analysis of the oscillation of the sacrum in

the vertical direction) and step width variability

(calculated by tracking the (x,y) coordinates of the

left and right foot) during the free walking segments.

The perpendicular distance between the left and

right foot along the walking path were used to

estimate the step width. Cognitive scores were

recorded separately for each condition. The average

and standard deviation across all conditions were

calculated. The scores were calculated based on the

correct responses given by the participants divided

by the total number of possible correct responses.

3 RESULTS

Four participants volunteered for the project.

Participant 3 and 4 had no history of concussion

while participant 1 had a mild concussion in the last

year. Participant 2 had suffered three concussions

and two within the last four years. The latest one, in

2012, occurred during a football game.

There were no significant differences in the

cognitive scores of the participants when it was

performed alone or when it was performed in

conjunction with a motor task given that they are

within one standard deviation of each other. The

average cognitive scores for the participants for the

N-2 cognitive task was 89.6% (std +/- 7.1%)

accuracy. Furthermore, the player who suffered a

concussion (participant 4) during the season did not

score. Similarly, there were no significant

differences in the motor performance of each

participant across the different segments of the

obstacle course in the motor-only condition given

that their trends where relatively constant and within

one standard deviation of each other. However,

when cognitive task and motor function were

simultaneously active during the obstacle course,

several kinematic performance variables emerge as

markers of concussion. Specifically, one participant

experienced a hard hit(s) during practice (Tuesday,

Fig.2). We proceeded to test him on the following

day. Subsequently, he played a regular season game

on the following Friday where he reported being

light headed, with headache, and felt nauseous. The

player was pulled from the game and was diagnosed

with a concussion by the athletic trainer. The

concussed participant was evaluated using a SCAT3

test a day after the game. He exhibited 16 of 22

possible symptoms and his Standardized Assessment

Concussion score dropped 9 points from his baseline

(out of a possible 25). He underwent concussion

protocol before returning to football activities two

weeks later.

The players who were not diagnosed with a

concussion showed a steady walking time during

free walking (Figure 2). While there were

differences between the non-concussed participants,

their relative trends remained flat throughout the

season. Furthermore, their values were close to the

99% confidence interval. However, the concussed

participant showed more variable walking time

throughout the season. For the test one day after

impact during practice, there was an increase

walking time during the free walking segment.

While the concussed participant showed a return to

baseline after the post-concussion protocol, he still

showed a variable trend post-concussion.

Figure 2: Walking time during free walking segment and

the 99% confidence interval band for each testing week

during cognitive + motor dual task.

Similar kinematic behaviours were also observed

in the step width variability (Figure 3). The step

width variability was normalized to height of each

participant for comparison. The non-concussed

group showed a relative steady trend over the course

of the season while the concussed participant who

showed an increasing trend in step width variability

after the impact during practice and the subsequent

concussion diagnosed during the game. However,

this increase in the relative trend of step width

variability was only observed when the participant

walk in the counter clockwise direction.

icSPORTS 2015 - International Congress on Sport Sciences Research and Technology Support

Figure 3: Step width variability during free walking

segment normalized with the height of each participant

and the 99% confidence interval band for each testing

week during cognitive + motor dual task.

Spectral analysis of the sacrum during free

walking also showed a shift in the median frequency

of the concussed participants after a head impact

during practice (Figure 4). Following the prescribed

post-concussion protocol, the median frequency

returned to baseline. However, the median frequency

of participant 2 deviated from the baseline behaviour

during the last week of testing while the other non-

concussed participants remained relatively steady

throughout the season.

Figure 4: Median frequency of the sacrum in the z-

direction during free walking segment and the 99%

confidence interval band for each testing week.

4 DISCUSSION

Concussion is difficult to objectively quantify.

Current testing mechanisms can be lengthy,

cumbersome and expensive. Concussions are often

not diagnosed until after gross cognitive and motor

dysfunctions are visibly observed. We demonstrated

here that our automated motor and cognitive test was

able to detect changes in performance before the

health staff of the football team detected the problem

in one player. The majority of concussions do not

exhibit any of these gross motor dysfunctions. This

is why there is a need for long-term monitoring of

youth athletes who are at risk of concussion. In this

paper, we proposed a kinematic-based testing

protocol using markerless motion capture system to

monitor the cognitive and motor performance of

football players throughout the playing season. This

system would reduce the amount of time required to

test each participant. Furthermore, it has the

potential to reveal cognitive and motor deficits that

are otherwise not visually observable.

Evidence suggests that youth athletes cognitively

recover more slowly than a college-age group (Field

et al., 2003, Covassin et al., 2012) and their

symptoms tend to last longer (Castile et al., 2012).

Furthermore, prolonged exposure to repetitive hits

could expedite the aging of the brain that could

result in more serious neurodegenerative diseases

such as chronic traumatic encephalopathy (CTE)

(Stern et al., 2011). Therefore, it is imperative that

preventive care is provided at any sign of motor and

cognitive deficiency to minimize the long-term

effects of these impacts. Objectively quantifying

these motor behaviours will remove the subjective

nature of concussion detection and management.

Castile (2012) found that 15.2% of youth athletes

who suffered concussion returned to play while still

symptomatic. This increases their exposure to

recurrent concussion and catastrophic effects of

second impact syndrome (Bowen, 2003, Bey and

Ostick, 2009, Cantu and Gean, 2010, McCrory et al.,

2012). Our approach will provide trainers with

continuous monitoring of their behaviour to ensure

that they are fully recovered from the effects of

concussion before return to play.

Under reporting is a major problem in the

detection and management of concussion. This could

be attributed to many factors such as the effect of

wanting more playing time and a perceived

weakness when not playing through injury. Talavage

(2014) longitudinal study of football player from

pre- to post-season found a high number of players

who were not diagnosed with concussion yet showed

measurable cognitive impairments. This

demonstrated that traditional cognitive testing might

not be sufficient to detect athletes who exhibit signs

of neurocognitive damage, yet never clinically

diagnosed with concussion. For example, participant

2 was never diagnosed with concussion during the

season; however, the median frequency was shifted

toward a lower frequency with a larger variation.

Relative to his baseline behaviour over the course of

the season, this player could be experiencing an

mTBI during the last week of the season. While

more investigative work is needed to determine the

reasons for the shift in the median frequency, the

Longitudinal Study on the Detection and Evaluation of Onset Mild Traumatic Brain Injury during Dual Motor and Cognitive Tasks

results should at least raise warning flags to medical

professionals to implement preventive measures to

ensure that the player is physically and cognitively

healthy given his history of concussion. The aim of

these preventive measures is to reduce the amount of

undetected and unreported incidences of concussion.

For youth athletes, these undetected injuries could

have a detrimental impact on their cognitive

development.

Long-term monitoring of these players is crucial

in identifying players who might exhibit signs of

cognitive and motor deficits without being clinically

diagnosed with concussion. Coaches and medical

professionals at the youth level often lack the

resources and training to deal with the complexity of

concussion detection since symptoms do not

manifest themselves unless major injury occurs.

Therefore, it is essential to provide these

professionals with simple metrics to easily monitor

the change in the motor and cognitive dysfunction of

these players throughout the season. Multi-tasking is

an essential element of sports; therefore, it is

imperative to monitor players’ performance when

subjecting to these types of tasks. In this study, we

aim to develop a testing protocol using a kinematic-

based dual-task paradigm to quickly and non-

invasively monitor high-risk players throughout the

season to track their motor and cognitive functions.

The ultimate goal would be to build a cognitive and

biomechanical passport for each player using data

collected longitudinally with a protocol such as the

one presented here. Then, the performance of each

athlete could be compared with that of his/her peers,

but also with that of him/herself. Any abrupt

changes could be considered as tell-tale signs that

mTBI has occurred. Since the test is also performed

without the cognitive tasks, the motor development

or alterations could be taken into consideration in

the detection of possible mTBI.

While the sample size is a limiting factor in this

study, we hope to test and monitor more players pre

and in-season to develop more comprehensive

metrics to measure the motor and cognitive function

of these at-risk athletes in future studies. This could

include simple performance metrics, such as the

ones presented here, but also full-body pattern

recognitions using more sophisticated techniques

such as machine learning, neural or Bayesian

network. Furthermore, as we are developing the

algorithms to detect mTBI, we are also exploring

hardware alternatives that could capture full-body

kinematics in a markerless fashion easily, and with

lower costs, so that such a system could eventually

become accessible to teams. Still, the present results

are encouraging, and are in line with previous

research using more conventional (yet more

complicated and time consuming) approaches

(Cossette et al., 2014). We hope to provide coaches

and medical professionals with a more simple, fast

and fully automated tool to objectively identify at-

risk players, and provide preventive care so as to

avoid the unhealthy consequences of multiple

undetected concussions and their impact long-term

health.

5 CONCLUSIONS

An automated kinematic-based dual-task paradigm

provided a promising testing mechanism to detect

the onset of concussion. This novel approach allows

for the monitoring of players over the entire season

and provides faster assessment of concussion effects

than traditional neurocognitive tests.

REFERENCES

Auerbach, P. S., Baine, J. G., Schott, M. L., Greenhaw, A.,

Acharya, M. G. & Smith, W. S. (2015). Detection of

concussion using cranial accelerometry. Clin J Sport

Med, 25, 126-32.

Bey, T. & Ostick, B. (2009). Second impact syndrome.

West J Emerg Med, 10, 6-10.

Bowen, A. P. (2003). Second impact syndrome: a rare,

catastrophic, preventable complication of concussion

in young athletes. J Emerg Nurs, 29, 287-9.

Cantu, R. C. & Gean, A. D. (2010). Second-impact

syndrome and a small subdural hematoma: an

uncommon catastrophic result of repetitive head injury

with a characteristic imaging appearance. J

Neurotrauma, 27, 1557-64.

Castile, L., Collins, C. L., McIlvain, N. M. & Comstock,

R. D. (2012). The epidemiology of new versus

recurrent sports concussions among high school

athletes, 2005-2010. Br J Sports Med, 46, 603-10.

Cossette, I., Ouellet, M. C. & McFadyen, B. J. (2014). A

preliminary study to identify locomotor-cognitive dual

tasks that reveal persistent executive dysfunction after

mild traumatic brain injury. Arch Phys Med Rehabil,

95, 1594-7.

Covassin, T., Elbin, R. J., Harris, W., Parker, T. & Kontos,

A. (2012). The role of age and sex in symptoms,

neurocognitive performance, and postural stability in

athletes after concussion. Am J Sports Med, 40, 1303-

12.

Fait, P., Swaine, B., Cantin, J. F., Leblond, J. &

McFadyen, B. J. (2013). Altered integrated locomotor

and cognitive function in elite athletes 30 days

postconcussion: a preliminary study. J Head Trauma

Rehabil, 28, 293-301.

icSPORTS 2015 - International Congress on Sport Sciences Research and Technology Support

Field, M., Collins, M. W., Lovell, M. R. & Maroon, J.

(2003). Does age play a role in recovery from sports-

related concussion? A comparison of high school and

collegiate athletes. J Pediatr, 142, 546-53.

Grady, M. F. (2010). Concussion in the adolescent athlete.

Curr Probl Pediatr Adolesc Health Care, 40, 154-69.

Greenwald, R. M., Gwin, J. T., Chu, J. J. & Crisco, J. J.

(2008). Head impact severity measures for evaluating

mild traumatic brain injury risk exposure.

Neurosurgery, 62, 789-98; discussion 798.

Khurana, V. G. & Kaye, A. H. (2012). An overview of

concussion in sport. J Clin Neurosci, 19, 1-11.

Kirchner, W. K. (1958). Age differences in short-term

retention of rapidly changing information. J Exp

Psychol, 55, 352-8.

Lee, H., Sullivan, S. J. & Schneiders, A. G. (2013). The

use of the dual-task paradigm in detecting gait

performance deficits following a sports-related

concussion: a systematic review and meta-analysis. J

Sci Med Sport, 16, 2-7.

McCrory, P., Davis, G. & Makdissi, M. (2012). Second

impact syndrome or cerebral swelling after sporting

head injury. Curr Sports Med Rep, 11, 21-3.

Parker, T. M., Osternig, L. R., Lee, H. J., Donkelaar, P. &

Chou, L. S. (2005). The effect of divided attention on

gait stability following concussion. Clin Biomech

(Bristol, Avon), 20, 389-95.

Pilon, M. & Belson, K. 2013. Seau Had Brain Disease

Found in Other Ex-Players. New York Times, January

11, 2013.

Post, A. & Blaine Hoshizaki, T. (2015). Rotational

acceleration, brain tissue strain, and the relationship to

concussion. J Biomech Eng, 137.

Rosenthal, J. A., Foraker, R. E., Collins, C. L. &

Comstock, R. D. (2014). National High School Athlete

Concussion Rates From 2005-2006 to 2011-2012. Am

J Sports Med, 42, 1710-1715.

Stern, R. A., Riley, D. O., Daneshvar, D. H., Nowinski, C.

J., Cantu, R. C. & McKee, A. C. (2011). Long-term

consequences of repetitive brain trauma: chronic

traumatic encephalopathy. PM R, 3, S460-7.

Talavage, T. M., Nauman, E. A., Breedlove, E. L., Yoruk,

U., Dye, A. E., Morigaki, K. E., Feuer, H. & Leverenz,

L. J. (2014). Functionally-detected cognitive

impairment in high school football players without

clinically-diagnosed concussion. J Neurotrauma, 31,

327-38.

Longitudinal Study on the Detection and Evaluation of Onset Mild Traumatic Brain Injury during Dual Motor and Cognitive Tasks