Analysis of Smooth Pursuit Assessment in Virtual Reality and Concussion Detection using BiLSTM Prithul Sarker1 Khondker Fariha Hossain1 Isayas Berhe Adhanom1

2025-04-30 0 0 681.54KB 11 页 10玖币

侵权投诉

Analysis of Smooth Pursuit Assessment in Virtual

Reality and Concussion Detection using BiLSTM

Prithul Sarker1, Khondker Fariha Hossain1, Isayas Berhe Adhanom1,

Philip K Pavilionis2, Nicholas G. Murray2, and Alireza Tavakkoli1

1Department of Computer Science and Engineering, University of Nevada, Reno,

United States

2School of Public Health, University of Nevada, Reno, United States

prithulsarker@nevada.unr.edu

Abstract. The sport-related concussion (SRC) battery relies heavily

upon subjective symptom reporting in order to determine the diagnosis

of a concussion. Unfortunately, athletes with SRC may return-to-play

(RTP) too soon if they are untruthful of their symptoms. It is criti-

cal to provide accurate assessments that can overcome underreporting

to prevent further injury. To lower the risk of injury, a more robust

and precise method for detecting concussion is needed to produce reli-

able and objective results. In this paper, we propose a novel approach

to detect SRC using long short-term memory (LSTM) recurrent neural

network (RNN) architectures from oculomotor data. In particular, we

propose a new error metric that incorporates mean squared error in dif-

ferent proportions. The experimental results on the smooth pursuit test

of the VR-VOMS dataset suggest that the proposed approach can pre-

dict concussion symptoms with higher accuracy compared to symptom

provocation on the vestibular ocular motor screening (VOMS).

Keywords: Concussion ·VOMS ·Smooth Pursuit ·Virtual Reality ·

LSTM.

1 Introduction

Mild Traumatic Brain Injuries (mTBI), also known as concussions, remain an

active major public health issue that aﬀects all levels of participation in sport

and recreational activities with an average of 1.6 to 3.8 million occurrences each

year [1]. A higher risk of traumatic brain injury (TBI) is linked with many of

these activities. In particular TBI aﬀects an estimated 1.7 million people every

year in the United States according to the estimation of the US Centers for

Disease Control and Prevention (CDC) [1]. Sport-related concussions (SRCs)

are a variety of injuries that produce transitory neurological impairment and

usually go away within two to four weeks [2].

Assessment of concussion is challenging as it requires recognizing the post-

injury symptoms rapidly. Some of the most validated and utilized sports con-

cussion assessment tools are Sport Concussion Assessment Tool 5 (SCAT5) [3],

arXiv:2210.11238v1 [physics.med-ph] 12 Oct 2022

2 Sarker et al.

Standard Assessment of Concussion (SAC) [4], King-Devick test (KDT) [5], Bal-

ance Error Scoring System (BESS) [6], Vestibular/Ocular Motor Screening Tool

(VOMS) [7], Blood-based biomarkers [8] etc. The tests often rely on the athlete’s

self-reported symptoms, and sometimes the symptoms go unrecognized for the

clinician perspective, resulting in misleading outcomes. Additionally, the physi-

cians’ concussion assessment system is sometimes inﬂuenced by self-reported bias

[9]. Even the most extensively used tests to detect concussion such as SCAT5

are not always accurate [10].

An immediate and transient loss of consciousness at the instant of head

trauma and returning to alertness within an hour or two is the main symptom of

concussion. Two of the commonly reported trajectories after SRC are 1) vestibu-

lar with dizziness, nausea, and vertigo as common symptoms and 2) ocular with

convergence insuﬃciency, blur, diplopia and/or headaches as impairments [2].

About 30% of concussed athletes have reported visual impairments during the

ﬁrst week of the injury [7]. The detection of vestibular and visual impairments

is the component of VOMS assessment to concussion evaluation and manage-

ment. The VOMS study contains ﬁve concise evaluations: (1) smooth pursuit,

(2) horizontal and vertical saccades, (3) convergence, (4) vestibular ocular reﬂex

(VOR), and (5) visual motion sensitivity (VMS). The test provokes the systems

responsible for controlling balance, vision, and movement, which help identify

issues that are not found in other assessments [7].

In our study, we used virtual reality (VR) headset to develop diﬀerent do-

mains of the assessment to eliminate subjectiveness and to introduce objec-

tiveness in the VOMS assessment. Smooth pursuit of the VOMS assessment is

performed to examine if there is a central pathology that prevents the eyes from

tracking moving targets.

2 Related Work

Concussion diagnosis is primarily justiﬁed in order to reduce the risk of imme-

diate or delayed harm following an external force head injury. As a result, tests

must be able to spot deviations from normal performance, behavior, and symp-

toms as well as changes or risks. The Standard Assessment of Concussion (SAC),

developed in 1997, comprises measures Orientation, short-term memory, atten-

tion, and delayed recall are the four components of the Standard Assessment

of Concussion (SAC), which assesses mental health [4]. Despite being a part of

the ﬁrst concussion evaluation, the SAC is insuﬃcient on its own to conduct a

complete concussion evaluation. The SAC score cannot be used to determine the

severity of a concussion or to support a player’s return to play on its own [11].

The Balance Error Scoring System (BESS) was developed by Guskiewicz et

al. to assess an athlete’s balance after a concussion [6]. Trials for the BESS test

must be completed by the athlete. A sequence of six balancing poses is used by

BESS to test postural stability by subjectively counting mistakes. The number

of errors on each trial determines the BESS score [6]. Although numerous studies

now conclude that there are postural and motor signs and symptoms that do not

Analysis of Smooth Pursuit Assessment 3

go away after a week or two, the BESS still has a tendency to be most sensitive

during the very acute period of recovery [12][11].

The Sideline Concussion Assessment Tool (SCAT) is the most frequently uti-

lized sports tool in the world. The SCAT1 was developed in 2004 by combining

the SAC, the Post-Concussion Symptom Scale (PCSS), sports-speciﬁc orienta-

tion questions, on-ﬁeld concussion indications, and RTP recommendations [3].

After its appearance, SCAT has been redesigned and updated its versions by

combining other assessments to improve the eﬃciency of the assessment. To

assist in the diagnosis of SRC, SCAT5 was created for qualiﬁed healthcare prac-

titioners, usually including doctors and sports trainers. The SCAT5, which has

undergone multiple signiﬁcant revisions, is still the industry standard for sideline

evaluation. Depending on the degree of the brain damage, administering SCAT5

might take as long as 20 minutes [3].

The above-described exams are extremely subjective in nature, and they take

a long time. Other tests have been tried and shown to be helpful in identifying

SRC on the sidelines at the time of injury in addition to the SCAT. The King-

Devick Test (KD) detects acutely subtle visual scanning abnormalities and is

based on the idea that mTBIs are a multi-system lesion. KD is a quick visual

scanning test lasting 2 to 3 minutes in which participants read numbers from

left to right for several lines on stimulation pages [5]. KD is a potential indica-

tor of mTBIs with good speciﬁcity and sensitivity levels. Combining KD with

other popular rapid sideline assessments of focus, memory, and balance increases

sensitivity and speciﬁcity [13].

Considering the other assessment methods, the VOMS has proven to be a

clinically eﬀective tool to diagnose concussion that recorded a high accuracy

of 89% with areas under the receiver operating characteristic curve (AUC) for

adolescent and collegiate-athlete populations [2]. Even VOMS manifested high

internal consistency in civilian [14] and military [2] people, making it one of

the most popular multi-domain assessments for the concussion detection sys-

tem. Current sideline assessment tools SCAT3, SAC, KDT, BESS, etc. failed

to inscribe vestibular and ocular system dysfunction comprehensively [15]. The

VOMS allows incorporating the missing ﬁndings related to the vestibular and

ocular system while assessing concussion. However, the VOMS assessment is also

distinctly subjective as the patients verbally rate changes in diﬀerent symptoms

including headache, dizziness, nausea, and fogginess on a scale of 0 to 10 after

each VOMS assessment in comparison to their pre-assessment state [7] [16].

Because the VOMS relies on subjective reporting of provoked symptoms, it

is important to understand how the eye movements may be related to these

provoked symptoms. In order to implement the concept and eliminate the sub-

jectivity from the assessment, we developed a novel method for VOMS testing

utilizing virtual reality. In this paper, we analyze the smooth pursuit test of

VOMS data collected through VR technology using diﬀerent deep learning ar-

chitectures.

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

AnalysisofSmoothPursuitAssessmentinVirtualRealityandConcussionDetectionusingBiLSTMPrithulSarker1,KhondkerFarihaHossain1,IsayasBerheAdhanom1,PhilipKPavilionis2,NicholasG.Murray2,andAlirezaTavakkoli11DepartmentofComputerScienceandEngineering,UniversityofNevada,Reno,UnitedStates2SchoolofPublicHealth,Un...

收起<<

Analysis of Smooth Pursuit Assessment in Virtual Reality and Concussion Detection using BiLSTM Prithul Sarker1 Khondker Fariha Hossain1 Isayas Berhe Adhanom1.pdf

共11页,预览3页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

Analysis of Smooth Pursuit Assessment in Virtual Reality and Concussion Detection using BiLSTM Prithul Sarker1 Khondker Fariha Hossain1 Isayas Berhe Adhanom1

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: