View
1
Download
0
Category
Preview:
Citation preview
Carnevale et al. BMC Musculoskeletal Disorders (2019) 20:546 https://doi.org/10.1186/s12891-019-2930-4
RESEARCH ARTICLE Open Access
Wearable systems for shoulder kinematics
assessment: a systematic review
Arianna Carnevale1, Umile Giuseppe Longo1* , Emiliano Schena2, Carlo Massaroni2, Daniela Lo Presti2,Alessandra Berton1, Vincenzo Candela1 and Vincenzo Denaro1
Abstract
Background: Wearable sensors are acquiring more and more influence in diagnostic and rehabilitation field toassess motor abilities of people with neurological or musculoskeletal impairments. The aim of this systematicliterature review is to analyze the wearable systems for monitoring shoulder kinematics and their applicability inclinical settings and rehabilitation.
Methods: A comprehensive search of PubMed, Medline, Google Scholar and IEEE Xplore was performed and resultswere included up to July 2019. All studies concerning wearable sensors to assess shoulder kinematics wereretrieved.
Results: Seventy-three studies were included because they have fulfilled the inclusion criteria. The results showedthat magneto and/or inertial sensors are the most used. Wearable sensors measuring upper limb and/or shoulderkinematics have been proposed to be applied in patients with different pathological conditions such as stroke,multiple sclerosis, osteoarthritis, rotator cuff tear. Sensors placement and method of attachment were broadlyheterogeneous among the examined studies.
Conclusions: Wearable systems are a promising solution to provide quantitative and meaningful clinicalinformation about progress in a rehabilitation pathway and to extrapolate meaningful parameters in the diagnosisof shoulder pathologies. There is a strong need for development of this novel technologies which undeniablyserves in shoulder evaluation and therapy.
Keywords: Shoulder kinematics, Upper limb, Wearable system, Inertial sensors, Smart textile
BackgroundShoulder kinematics analysis is a booming researchfield due to the emergent need to improve diagnosisand rehabilitation procedures [1]. The shoulder com-plex is the human joint characterized by the greatestrange of motion (ROM) in the different planes ofspace.Commonly, several scales and tests are used to
evaluate shoulder function, e.g., the Constant-Murleyscore (CMS), the Simple Shoulder test (SST), theVisual Analogue Scale (VAS) and the Disability of the
© The Author(s). 2019 Open Access This articInternational License (http://creativecommonsreproduction in any medium, provided you gthe Creative Commons license, and indicate if(http://creativecommons.org/publicdomain/ze
* Correspondence: g.longo@unicampus.it1Department of Orthopaedic and Trauma Surgery, Campus Bio-MedicoUniversity, Via Álvaro del Portillo, 200, 00128 Rome, ItalyFull list of author information is available at the end of the article
Arm, Shoulder, and Hand (DASH) score [2–4]. How-ever, despite their easy-to-use and wide application inclinical settings, these scores conceal an intrinsic sub-jectivity [2–4], inaccuracy in approaching diagnosis,follow-up and treatment of the pathologies. Quantita-tive and objective analyses are rapidly developing as avalid alternative to evaluate shoulder activity level, togauge its functioning and to provide informationabout movement quality, e.g., velocity, amplitude andfrequency [5, 6]. This interest in the use of measuringsystems is growing in many medical fields to recordinformation of clinical relevance. For example, elec-tromyography (EMG), force sensors, inertial measure-ment units (IMU), accelerometers, fiber optic sensorsand strain sensors are employed for human motionanalysis, posture and physiological parameters moni-toring [7–10]. From a technological viewpoint, the
le is distributed under the terms of the Creative Commons Attribution 4.0.org/licenses/by/4.0/), which permits unrestricted use, distribution, andive appropriate credit to the original author(s) and the source, provide a link tochanges were made. The Creative Commons Public Domain Dedication waiverro/1.0/) applies to the data made available in this article, unless otherwise stated.
http://crossmark.crossref.org/dialog/?doi=10.1186/s12891-019-2930-4&domain=pdfhttp://orcid.org/0000-0003-4063-9821http://creativecommons.org/licenses/by/4.0/http://creativecommons.org/publicdomain/zero/1.0/mailto:g.longo@unicampus.it
Carnevale et al. BMC Musculoskeletal Disorders (2019) 20:546 Page 2 of 24
monitoring of shoulder motion is challenging due tothe complexity of joint kinematic which require thedevelopment of protocols exploiting sensing technol-ogy as much as possible reliable and unobtrusive. Inthe last years, a great number of human motion ana-lysis systems have been largely employed for objectivemonitoring. These systems can be classified into twomain categories: wearable and non-wearable [11]. Thelast one includes electromagnetic tracking systems(e.g., Fastrak) [12], ultrasound-based motion analysissystems (e.g., Zebris) [13], stereo-photogrammetricand optoelectronic systems (e.g., VICON, Optotrak,BTS SMART-D) often used as gold standard [14–17].These systems based on magnetic field, ultrasoundand cameras are effectively suitable for 3D motiontracking and analysis due to their accuracy, precisionand reliability [18]. On the other hand, such systemsrequire expensive equipment, frequent calibration and,overall, they restrict measurements in structured en-vironment [19]. Wearable systems overcome theseshortcomings and they are a promising solution forcontinuous and long-term monitoring of human mo-tion in daily living activities. Gathering data in un-structured environment continuously (e.g., homeenvironment) provide additional information com-pared to those obtainable inside a laboratory [20].Wearable sensor-based systems, intended for kine-
matics data extraction and analyses, are acquiringmore and more influence in diagnostic applications,rehabilitation follow-up, and treatments of neuro-logical and musculoskeletal disorders [21, 22]. Suchsystems comprise accelerometers, gyroscopes, IMU,among others [23]. Patients’ acceptance of monitoringsystems that should be worn for long-time relies onsensors’ features whose must be lightweight, unobtru-sive and user-friendly [24]. The increasing trend toadopt such wearable systems has been promoted bythe innovative technology of micro-electro-mechanicalsystems (MEMS). MEMS technology has fostered sen-sors’ miniaturization, paving the way for a revolution-ary technology suited to a wide range of applications,including extraction of clinical-relevant kinematics pa-rameters. In recent years, there has been growth inthe use of smart textile-based systems which integratesensing units directly into garments [11, 25, 26].Moreover, in the era of big data, machine learningtechnical analysis can improve home rehabilitationthanks to the recognition of the quality level of per-formed physical exercises and the possibility to pre-vent disorders in patients’ movement [27].The aim of this systematic literature review is to de-
scribe the wearable systems for monitoring shoulderkinematics. The authors want to summarize the mainfeatures of the current wearable systems and their
applicability in clinical settings and rehabilitation forshoulder kinematics assessment.
MethodsLiterature search strategy and study selection processA systematic review was executed applying thePRISMA (Preferred Reporting Items for SystematicReviews and Meta-Analyses) guidelines [28]. Full-textarticles and conference proceedings were selectedfrom a comprehensive search of PubMed, Medline,Google Scholar and IEEE Xplore databases. Thesearch strategy included free text terms and Mesh(Medical Subject Headings) terms, where suited.These terms were combined using logical Boolean op-erators. Keywords and their synonyms were combinedin each database as follows: (“shoulder biomechanics”OR “upper extremit*” OR “shoulder joint” OR “scapu-lar-humeral” OR “shoulder kinematics” OR “upperlimb”) AND (“wearable system*” OR “wearable de-vice*” OR “wearable technolog*” OR “wearable elec-tronic device*” OR “wireless sensor*” OR “sensorsystem” OR “textile” OR “electronic skin” OR “inertialsensor”). No filter was applied on the publication dateof the articles, and all results of each database wereincluded up to July 2019. After removal of duplicates,all articles were evaluated through a screening of titleand abstract by three independent reviewers. Thesame three reviewers performed an accurate readingof all full-text articles assessed for eligibility to thisstudy and they performed a collection of data tominimize the risk of bias. In case of disagreementamong investigators regarding the inclusion and ex-clusion criteria, the senior investigator made the finaldecision.Inclusion criteria were:
i) The studies concern wearable systems as a tool toassess upper limb kinematics;
ii) The studies used sensors directly stuck on thehuman skin by means of adhesive, embedded withinpockets, straps or integrated into fabrics;
iii) Systems intended for motion recognition andrehabilitation;
iv) Articles are written in English language;v) Papers are published in a peer-reviewed journal or
presented in a conference;
Exclusion criteria were:
i) Use of prosthetics, exoskeleton or robotic systems;ii) Wearable system not directly worn or tested on
human;iii) The study concerns wearable systems for full-body
motion tracking;
Carnevale et al. BMC Musculoskeletal Disorders (2019) 20:546 Page 3 of 24
iv) Shoulder joint is not included;v) Reviews, books.
Data extraction processData extraction was executed on 73 articles. Data was ex-tracted on the base of the following checklist: authors, yearand type of publication (i.e., conference or full-text); typ-ology, number, brand and placement of the sensors usedto measure or track the kinematic of the interested joint,wearability of the system, target parameters with regard tothe shoulder; system used as gold standard to assess thewearable systems’ performance; tasks executed in the as-sessment protocol; characteristics of the participants in-volved in the study and aim of the study.
ResultsThe literature search returned 1811 results and additional14 studies were identified through other sources. A totalof 73 studies fulfilled the inclusion criteria (Fig. 1), ofwhich 27% were published on conference proceedings andthe remaining 73% on peer-reviewed journal.Three levels of analysis have been emphasized in this
survey: A. application field and main aspects covered, B.
Fig. 1 PRISMA 2009 flow diagram
the typology of sensors exploited to measure kinematicparameters, C. the placement of the single measurementunits on the body segment of interest and how sensingmodules are integrated into the wearable system from awearability viewpoint.
Application fieldFifteen out of the 73 studies focused on evaluating upperlimbs motion in case of musculoskeletal diseases (e.g.,osteoarthritis, rotator cuff tear, frozen shoulder), 26 onneurological diseases and ap-plication in neurorehabilita-tion (e.g., stroke, multiple sclerosis), 15 on general rehabili-tation aspects (e.g., home rehabilitation, physiotherapymonitoring) and 17 focusing on validation and develop-ment of systems and algorithm for monitoring shoulderkinematics. Tables 1, 2, 3 and 4 include, for each of theidentified application fields, data listed in the previous dataextraction process section.
Sensing technologySome studies combined different sensors in their mea-surements system. The most used sensors are acceler-ometers, gyroscopes and magnetometers, a combination
Table
1Shou
lder
motionmon
itorin
gforapplicationin
patientswith
musculoskeletaldisorders
Reference,Year,
Type
ofpu
blication
Sensors,Brand
Placem
entandwearability
Target
shou
lder
parameters,Perfo
rmance
Gold
standard
Task
executed
Participants
Aim
Coley
2007,[29]
Full-Text
IMU(n=2),
AnalogDevice
(gyr:A
DXR
S250,acc:
ADXL
210)
Bilateral:
humeri(po
steriorly,
distally)
Adh
esivepatch
ShRO
M(HTjointangles),
RMSE
=5.81°
Mean=1.80°
Zebris
CMS-HS
ShFLX-EXT
ShAB-AD
ShIER
HS(n=10),
25.1±4.1Y
P(n=10),
7RC
,3OA,
4F,6M,
62.4±10.4Y
Find
objectivescores
toassess
shou
lder
functio
n;to
validate
such
scores
comparin
ghe
althy
andaffected
shou
lders
Coley
2008,[30]
Full-Text
IMU(n=4),
AnalogDevice
(gyr:A
DXR
S250,
acc:ADXL
210)
Bilateral:
humeri(po
steriorly,
distally),thorax
Adh
esivepatch
ShRO
M(HTjointangles)
–Dailyactivities
(8h)
HS(n=35)
32±8Y
Quantify
usageof
shou
lder
and
thecontrib
utionof
each
shou
lder
indaily
lifeactivities
Jolles2011,[31]
Full.Text
IMU(n=4),
AnalogDevice(gyr:
ADXR
S250,acc:
ADXL
210)
Bilateral:
Hum
eri(distally,
posteriorly),thorax
(×2)
Adh
esivepatch
ShRO
M(HTjointangles),
r=0.61–0.80
–7activities
ofSST
P(n=34)
27RC
,7OA
25M,9
F57.5±9.9
CG(n=31)
17M,14F
33.3±8.0
Validateclinicallyshou
lder
parametersin
patientsafter
surgeryforGHOAandRC
disease
Duc
2013,[5]
Full-Text
IMU(n=3),
AnalogDevice(gyr:
ADXR
S250,acc:
ADXL
210)
Bilateral:
Hum
eri(distal,p
osterio
r),sternu
mAdh
esivepatch
ShRO
M(HTjointangles),r2=
0.13-0.52(CMS)
r2=0.06-0.30(DASH
)r2=0.03-0.31(SST)
–Free
movem
ents,
daily
activities
HS(n=41)
34±9Y
P(n=21)
53±9Y
RC(unilateral)
Validateametho
dto
detect
movem
entof
thehu
merus
relativeto
thetrun
kandto
provideou
tcom
esparameters
aftershou
lder
surgery(freq
uency
ofarm
movem
entandvelocity)
Körver
2014,[32]
Full-Text
IMU(n=2),
Inertia-Link-2400-SKI,
MicroStrain
Bilateral:
Hum
erus
(distal,po
sterior)
Adh
esive
ShRO
M(HTjointangles),
r=0.39
(DASH
)r=
0.32
(SST)
–Handto
theback,
Handbe
hind
the
head
HS(n=100)
37M,63F
40.6±15.7Y
P(n=15)
5M,10F
57.7±10.4Y,
Subacrom
ial
imping
emen
t
Investigateabou
tthecorrelation
betw
eensubjective(clinicalscale)
andob
jective(IM
U)assessmen
tof
shou
lder
ROM
durin
galong
-term
perio
dof
follow-up
VanDen
Noo
rt2014,[33]
Full-Text
M-IM
U(n=4),
Xsen
sMTw
Unilateral:
Thorax,scapu
la,
UA,FA
Straps,skintape
ShRO
M(HTandST
jointangles)
–FLX(sagittalplane)
andAB(fron
talp
lane
)with
elbextend
edandthum
bpo
intin
gup
HS(n=20)
3M,17F
36±11
Y
Evaluate
theintra-
andinter-ob
server
reliabilityandprecisionof
3Dscapulakine
matics
Pichon
naz2015,
[34]
Full-Text
IMU(n=3),
AnalogDevice
(acc:A
DXL
210,
gyr:ADXR
S250)
Bilateral:
Sternu
m,H
umeri(po
sterior,
distal)
Skin
tape
ShRO
M(HTjointangles)
–Free
movem
ents
(7h)
HS(n=41)
23M,18F
34.1±8.8Y
P(n=21)
Exploredo
minantandno
n-do
minant
arm
usageas
anindicatorof
UL
functio
nafterrotatorcuffrepair
durin
gthefirstyear
aftersurgery
Carnevale et al. BMC Musculoskeletal Disorders (2019) 20:546 Page 4 of 24
Table
1Shou
lder
motionmon
itorin
gforapplicationin
patientswith
musculoskeletaldisorders(Con
tinued)
Reference,Year,
Type
ofpu
blication
Sensors,Brand
Placem
entandwearability
Target
shou
lder
parameters,Perfo
rmance
Gold
standard
Task
executed
Participants
Aim
14M,7
F53.3±9Y
RC
Roldán-Jim
énez
eCuesta-Vargas
2015,[35]
Full-Text
M-IM
U(n=4),
InterSen
seInertiaCub
e3
Unilateral:
Sternu
m,hum
erus,
scapula,FA
(nearwrist)
adhe
sive
tape
and
band
age
ShRO
M(HTandST
jointangles)
–180°
shABandEXTwith
wriin
neutralp
osition
andelbextend
ed
HS(n=11)
8M,3
F24.7±4.2Y
Analyse
ULangu
larmob
ility
and
linearacceleratio
nin
three
anatom
icalaxes
VanDen
Noo
rt2015,[36]
Full-Text
M-IM
U(n=4),
Xsen
sMTw
Unilateral:
Thorax,Scapu
la,U
A,FA
(ISEO
protocol
[33,37])
Straps,skintape
ShRO
M(STjointangles)
–Hum
eralFLX(sagittal
plane)
andAB(fron
tal
plane)
with
elbextend
edandthum
bpo
intin
gup
P(n=10)
8M,2
F24–63Y
SD
Evaluate
thechange
in3D
scapular
kine
maticsusingasing
leand
doub
leanatom
icalcalibratio
nwith
ascapular
locatorversus
standard
calibratio
n;evaluate
differencein
3Dscapular
kine
matics
betw
eenstaticpo
stureanddynamic
humeralelevation
Roldán-Jim
énez
eCuesta-Vargas
2016,[38]
Full-Text
M-IM
U(n=3),
InterSen
se,
InertiaCub
e3
Unilateral:
Hum
erus
(middlethird
,slightlypo
sterior),
scapula,
sternu
mDou
ble-side
dtape
,elastic
band
age
ShRO
M(GHandST
jointangles)
–Sh
AB(fron
talp
lane
)Sh
FLX(sagittalplane)
Wriin
neutralp
osition
andelbextend
ed
Youn
gHS(n=11),
18–35Y
3M,8
FAdu
ltHS(n=14)
>40
Y5M,9F
Analyze
differences
inshou
lder
kine
maticsin
term
sof
angu
lar
mob
ility
andlinearacceleratio
nrelatedto
age
Wang2017,[26]
Full-Text
M-IM
U(n=3),
Adafru
itFLORA
9-DOF
Unilateral:
Shou
lder
(flat
partof
acromion),spine
(C7-T1,
T4-T5)
Zipp
edvest,elasticstrap
with
Velcro
ShRO
M(STjointangles),
RMSE
~3.57°
PST-55/
110
series
60°sh
FLXwith
elb
extend
edandthum
bpo
intin
gup
,place
acookingpo
ton
ashelf,
40°sh
ELin
thescapular
planewith
elbextend
edandthum
bpo
intin
gup
,place
abo
ttleof
water
onashelf
Pwith
musculoskeletal
shou
lder
pain
(n=
8) 3M,5
F50
±6.44Y
Physiotherapist
(n=5)
Evaluate
usability
ofasm
art
garm
ent-supp
ortedpo
stural
feed
back
scapular
training
inpatientswith
musculoskeletal
shou
lder
pain
andin
physiotherapists
who
take
care
patientswith
shou
lder
disorders
Aslani2018,[39]
Full-Text
M-IM
U(n=1),Bosh
SensortecBN
O055
EMG,M
yoWare
MuscleSensor
Unilateral:
UA,d
eltoid
(2surface
electrod
eson
each
delto
idsection)
Band
ShRO
M(azimuthaland
elevationangles)
–Arm
EL(m
edially,anteriorly,
cranially,p
osterio
rly,laterally)
with
elbfully
extend
ed
HS(n=6)
4M,2
F27.3±3.4Y
P(n=1),M
Frozen
Sh42
Y
Evaluate
ameasuremen
tsprotocol
toassess
thepe
rform
ance
ofthe
shou
lder
bycombining
both
ROM
andelectrom
yography
measuremen
ts
Carbo
naro
2018,
[40]
M-IM
U(n=3),Xsens
MTw
Unilateral:
Sternu
m,FA(distal),
ShRO
M(GHandST
jointangles)
–Extra-rotatio
n,Arm
AB(with
different
load)
HS(n=5)
Testadigitalapp
licationintend
edfortele-m
onito
ringand
Carnevale et al. BMC Musculoskeletal Disorders (2019) 20:546 Page 5 of 24
Table
1Shou
lder
motionmon
itorin
gforapplicationin
patientswith
musculoskeletaldisorders(Con
tinued)
Reference,Year,
Type
ofpu
blication
Sensors,Brand
Placem
entandwearability
Target
shou
lder
parameters,Perfo
rmance
Gold
standard
Task
executed
Participants
Aim
Con
ference
scapula(sim
ilarto
the
configurationin
[37])
Elastic
band
s
tele-reh
abilitatio
nof
theshou
lder
muscular-skeletaldiseases
Hurd2018,[41]
Full-Text
Acc
buil.in
ActiGraph
(n=2),A
ctiGraph
GT3XP
-BLE
Unilateral:
Wrist,UA(m
id-bicep
slevel)
Velcro
strap
ShRO
M–
ADL
P(n=14)
7M,7
F73
±6Y
GHOA,RCdisease
Evaluate
change
sin
pain,
self-repo
rted
functio
nandob
jective
measuremen
tof
uppe
rlim
bactivity
afterRSA
Lang
ohr2018,[6]
Full-Text
IMU(n=5),YEI
Techno
logy
Bilateral:
Sternu
m,U
A(lateral
aspe
ctsof
the
midhu
merus),FA
(dorsalaspectof
the
wrist)
Shirt
ShRO
M(hum
eral
elevationandplaneof
elevationangles)
–Dailyactivities
(1day,mon
itorin
gof
11±3
h)
P(n=36)
73±10
YTSA,RTSA
Determinethetotald
ailyshou
lder
motionof
patientsafterTSAand
RTSA
,com
pare
themotionof
the
arthroplasty
shou
lder
with
that
ofthecontralateralasymptom
atic
jointandcompare
thedaily
motionof
TSAandRTSA
shou
lders
accaccelerometer,g
yrgy
roscop
e,mag
nmag
netometer,IMUInertia
lMeasuremen
tUnit,M-IM
UMag
neto
andInertia
lMeasuremen
tUnit,UAUpp
erArm
,FAFo
rearm,R
OM
Rang
eof
motion,
HThu
merotho
racic,ST
scap
ulotho
racic,GHglen
ohum
eral,Shshou
lder,w
riwrist,elbelbo
w,FLX-EXT
flexion
-exten
sion
,AB-ADab
duction-ad
duction,
IERinternal-externa
lrotation,
RMSE
root
meansqua
reerror,r=
correlation,
r2,coe
fficient
ofde
term
ination,
HSHealth
ysubject,CG
Con
trol
Group
,Ppa
tient,M
male,
Ffemale,
RCRo
tatorCuff,OAOsteo
arthritis,Y
Yearsold,
SSTSimpleSh
oulder
test,D
ASH
Disab
ility
oftheArm
,Sho
ulde
ran
dHan
d,CM
SCon
stan
tMurleyScore
Carnevale et al. BMC Musculoskeletal Disorders (2019) 20:546 Page 6 of 24
Table
2Shou
lder
motionmon
itorin
gforapplicationin
patientswith
neurolog
icaldisorders
Reference,Year,
Type
ofpu
blication
Sensors,Brand
Placem
entand
wearability
Target
shou
lder
parameters,
Perfo
rmance
Goldstandard
Task
executed
Participants
Aim
Bartalesi2005,
[8]
Con
ference
Strain
sensor,
WACK
ERLtd.
(ELA
STOSILLR
3162
A/B)
Unilateral:
Sensors’segm
ent
seriesalon
grig
htup
perlim
bShirt
ShRO
M–
–HS
Wearablegarm
entableto
reconstruct
shou
lder,w
ristandelbo
wmovem
ent
tocorrectstroke
patients’rehabilitation
exercises
Hester2006,
[42]
Con
ference
Acc
(n=6)
Unilateral:
thum
b,inde
x,back
ofthehand
,FA
andUA(m
edially),
thorax
Adh
esive
ShRO
M–
Reaching
,prehe
nsion,
manipulation
SP(n=12)
Pred
ictclinicalscores
ofstroke
patients’
motor
abilities
Zhou
2006,[43]
Full-Text
IMU(n=2),
Xsen
sMT9B
Unilateral:
Wrist(inwards),
elbo
w(outwards)
-
Shorientationand
positio
n–
FAFLX-EXT
FAPR-SU,reach
test
HS(n=1,
M)
Prop
oseadata
fusion
algo
rithm
tolocate
theshou
lder
jointwith
outdrift
Willmann2007,
[44]
Con
ference
M-IM
U(n=4),
Philips
Unilateral:
Torso,shou
lder,
UA,FA
Garmen
t
ShRO
M–
–HS
Provideaho
mesystem
forup
per
limbrehabilitationin
stroke
patients
Zhou
2008,[45]
Full-Text
M-IM
U(n=2),
Xsen
sMT9B
Unilateral:
FA(distally,near
thewristcenter),
UA(laterally,on
thelinebe
tween
thelateralepicond
yle
andtheacromion
process)
Velcro
straps
Shorientationand
positio
n,RM
SE=2.5°-4.8°
CODA
reaching
shrugg
ing
FArotatio
n
HS(n=4,
M)
20–40Y
Validatedata
fusion
algo
rithm
Giorgino2009,
[46]
Full-Text
Strain
sensor
(n=19),
WACK
ERLtd.
(ELA
STOSIL
LR3162
A/B)
Unilateral:
Sensors’sensing
segm
entsdistrib
uted
over
theUA,FA,
shou
lder,elbow
,wrist
Shirt
ShRO
M–
GHFLX(sagittalplane)
LateralA
BER
HS(n=1)
Describeasensinggarm
entfor
posturerecogn
ition
inne
urolog
ical
rehabilitation
Lee2010,[47]
Full-Text
Acc
(n=2),
Freescale
MMA7261
QT
Unilateral:
UA,FA
Velcro
strap
ShRO
M,
Meanerror~0°-3.5°
gon
FLX-EXT(sagittalplane)
HS(n=1)
Validatepe
rform
ance
andaccuracy
ofthesystem
Che
eKian
2010,
[48]
Con
ference
OLE
(n=1)
Acc
(n=1)
Unilateral:
UA,elbow
Adh
esivepatch
ShRO
MIGS-190
Cyclic
movem
entswith
arm
exerciser
HS(n=1)
Validatefeasibility
andpe
rform
ance
(accuracy,repe
atability)of
theprop
osed
sensingsystem
design
edto
assiststroke
patientsin
uppe
rlim
bho
merehabilitation
Patel2010,[49]
Full-Text
Acc
(n=6)
Unilateral:
thum
b,inde
x,back
ShRO
M–
8activities
ofFA
SSP
(n=24)
57.5±11.8
Evaluate
accuracy
ofFA
Sscoreob
tained
via
analysisof
theaccelerometersdata
Carnevale et al. BMC Musculoskeletal Disorders (2019) 20:546 Page 7 of 24
Table
2Shou
lder
motionmon
itorin
gforapplicationin
patientswith
neurolog
icaldisorders(Con
tinued)
Reference,Year,
Type
ofpu
blication
Sensors,Brand
Placem
entand
wearability
Target
shou
lder
parameters,
Perfo
rmance
Goldstandard
Task
executed
Participants
Aim
ofthehand
,FA,U
A,
trun
k-
Ycomparin
gsuch
estim
ates
with
scores
provided
byan
expe
rtclinicianusingthis
scale
Pérez2010,[15]
Full-Text
M-IM
U(n=4),
Xsen
sMTi
Unilateral:
UA(18cm
from
acromion),FA
(25cm
from
epicon
dyle),hand
(5.5cm
from
distal
radio-cubitaljoint),
back
(position
isno
trelevant)
Strap
ShRO
M,
r=0.997(fo
rsh
IR,after
calibratio
n)
BTSSM
ART-D
ShFLX-EXT
Shho
rizon
talA
B-AD
ShIR
ElbFLX
ElbPR-SU
WriFLX-EXT
Servingwater
from
ajar
HS(n=1,F)
Validatethesystem
Bento2011,[50]
Con
ference
M-IM
U(n=4)
Bilateral:
Shou
lder,U
A,w
rist
(affected
and
unaffected
side
)Strap
Shorientation
andpo
sitio
n–
FAto
Table
FAto
box
Extend
elbo
wHandto
table
Handto
box
SP(n=5,
M)
35–73Y
Prelim
inaryvalidationof
asystem
ableto
quantifyup
perlim
bmotor
functio
nin
patientsafterne
urolog
icaltrauma
Ngu
yen2011,
[51]
Full-Text
OLE
(n=3)
Acc
(n=3)
Unilateral:
Shou
lder,elbow
,wrist
Clothingmod
ule
fixed
with
Velcro
straps
ShRO
M,
Test2:RM
SE=3.8°
(gon
),RM
SE=3.1°
(SW)
Test3:ICC=0.975(sh)
Test2:go
n,Shape-
Wrap
Test2:Bend
and
Flex
elbo
wTest3:reaching
task
Test2:
HS(n=3,
M);
Test3:
HS(n=5)
Validationof
theprop
osed
motion
capturesystem
Ding2013,[52]
Full-Text
M-IM
U(n=2),
AnalogDevice
(acc:A
DXL320),
Hon
eyWell(magn:
HMC1053),Silicon
SensingSystem
(gyr)
Unilateral:
UA(distal,ne
arelbo
w),FA
(distal,
near
wrist)
Velcro
strap
ShRO
M–
Replicationof
10reference
arm
posture
HS(n=5)
20–27Y
Che
ckthefeasibility
oftheprop
osed
system
which
measuresorientationand
correctsup
perlim
bpo
stureusing
vibrotactileactuators
Lee2014,[53]
Full-Text
M-IM
U(n=7),
AnalogDevice
(acc:ADXL
345)
InvenSen
se(gyr:ITG
3200)
Hon
eywell
(magn:HMC5883L)
Bilateral:
Back,U
A,FAand
hand
Strap
ShRO
M,
Test1:r=
0.963
Test2:RM
SE<5°
Test1:
gon
Test2:VICON
ShFLX-EXT
ShAB
Arm
IER
ElbFLX
FAPR-SU
WriFLX-EXT
Wriradial-ulnar
deviation
SP(n=5)
2M,3
FMeanage:
68Y
Introd
uceasm
artpho
necentric
wireless
wearablesystem
ableto
automatejoint
ROM
measuremen
tsandde
tect
thetype
ofactivities
instroke
patients
Bai2015,[54]
Full-Text
M-IM
U(n=4or
n=1),
Xsen
sMTx
Unilateral
Scapula,UA,FA,
back
ofthehand
oron
lyon
UA
Velfoam
,Velcro
straps
ShRO
M(upp
erlim
bsegm
ents
orientationandpo
sitio
n)
Test1:
gon
Test2:
gon,VICO
N
ShFLX-EXT
ShIER
ShAB-AD
ElbFLX-EXT
FAPR-SU
WriFLX-EXT
Wriradial-ulnar
deviation
HS(n=10)
8M,2
F20–38Y
P(n=1),F
41Y
Evaluate
afour-sen
sorsystem
anda
one-sensor
system
,investig
atewhe
ther
thesesystem
sareableto
obtain
quantitativemotioninform
ation
from
patients’assessmen
tdu
ring
neuroreh
abilitatio
n
Carnevale et al. BMC Musculoskeletal Disorders (2019) 20:546 Page 8 of 24
Table
2Shou
lder
motionmon
itorin
gforapplicationin
patientswith
neurolog
icaldisorders(Con
tinued)
Reference,Year,
Type
ofpu
blication
Sensors,Brand
Placem
entand
wearability
Target
shou
lder
parameters,
Perfo
rmance
Goldstandard
Task
executed
Participants
Aim
Ertzgaard2016,
[55]
Full-Text
IMU(n=5),
AnalogDevice,
Adis16,350
Bilateral:
Back
(upp
erbo
dy),
UAandFA
Strap
ShRO
M(HTjointangles),
ICC=0.768–0.985
Cod
aMovem
entsthat
mim
icactivities
ofdaily
life
HS(n=10)
2M,8
F34.3±13.1Y
SP(n=1),
F,43
Y
Validationstud
yto
characterizeelbo
wandshou
lder
motiondu
ring
functio
naltaskusingamod
ified
Expo
sure
Variatio
nAnalysis(EVA
)
Lorussi2016,
[56]
Full-Text
M-IM
U(n=2),
Xsen
sMTw
Strain
sensor
(n=
1),
Smartex
Unilateral:
M-IM
Usensorson
sternu
mandUA
Textile-based
strain
sensor
ontheback
(from
thespineto
scapula)
Shirt
ShRO
M(HTjointangleand
scapular
translation)
BTSSM
ART-DX
UAFLX(sagittalplane)
UAAB(fron
talp
lane
)HS(n=5)
Validatesensorsanddata
fusion
algo
rithm
toreconstructscapular-hum
eral
rhythm
Mazam
enos
2016,[57]
Full-Text
M-IM
U(n=2),
Shim
mer2r
Unilateral:
UA(distal,ne
arelbo
w),FA
(distal,
near
wrist)
Straps
ShRO
M(seg
men
ts’
orientationand
positio
n)
–EXP1:Reach
andretrieve,lift
object
tomou
th,rotatean
object
EXP2:p
reparin
gacupof
tea
HS(n=18)
M,F
25–50Y
SP(n=4)
M,F
45–73Y
Evaluate
thepe
rform
ance
androbu
stne
ssof
ade
tectionanddiscrim
ination
algo
rithm
ofarm
movem
ents
Jiang
2017,[58]
Con
ference
Acc
(n=4),
AnalogDevice
ADXL362
EMG(n=4),A
nalog
DeviceAD8232
Tempe
rature
(n=1)
Unilateral:
alon
gup
per
limb
Shirt
ShRO
M–
4typicaljoint
actio
nspe
rform
edin
clinical
assessmen
t
HS(n=1),
MIntrod
ucean
IoT-Basesup
perlim
brehabilitationassessmen
tsystem
for
stroke
patients
Li2017,[59]
Full-Text
IMU(n=2),
InvenSen
se,
MPU
-9250
EMG(n=10),
American
Imex,
Dermatrode
Unilateral:
IMU:U
A,w
rist
EMG:FA(n=8),
UA(n=2)
Stretchablebe
lt
ShRO
M–
11tasksinclud
ing:
ShFLX,
ShAB
WriFLX-EXT,
Fetchandho
ldaballor
acylindricroll,finge
rto
nose,
touchtheback
ofthesh,FA
PR-SU
HS(n=16)
10M,6
F36.25±
15.19Y
SP(n=18)
11M,7
F55.28±
12.25Y
Prop
osedata
fusion
from
IMUand
surface
EMGforqu
antitativemotor
functio
nevaluatio
nin
stroke
subjects
New
man
2017,
[60]
Full-Text
IMU(n=3),G
ait
UpSA
Physilog4
Bilateral:
sternu
m,U
A(posterio
r)Velcro
straps,
adhe
sive
patch
ShRO
M(HTjointangles)
–Reaching
movem
ents(lateral,
forw
ard,
upward)
Children
(n=30)
10.6±3.4Y
17bo
ys,13
girls
Testan
IMU-based
system
tomeasure
uppe
rlim
bfunctio
nin
childrenwith
hemiparesisandits
correlationwith
clinicalscores
Yang
eTan
2017,[61]
Con
ference
M-IM
U(n=4),
APD
MOpal
Unilateral:
Waist,tho
rax,UA
(distal,ne
arelbo
w),
FA(distal,ne
arelbo
w)
Straps
ShRO
M(HTjointangles)
Optitrack
Movem
entsrelatedwith
waistjoint,sh
jointandelb
jointsto
achievejoint
rotatio
n
HS(n=2)
Validatetheprop
osed
motion
tracking
system
Dauno
ravicien
eM-IM
U(n=6),
Bilateral:
ShRO
M–
FNTtest(ShEXT,sh
AB,elb
CG(n=24)
TestaM-IM
Ubasedsystem
to
Carnevale et al. BMC Musculoskeletal Disorders (2019) 20:546 Page 9 of 24
Table
2Shou
lder
motionmon
itorin
gforapplicationin
patientswith
neurolog
icaldisorders(Con
tinued)
Reference,Year,
Type
ofpu
blication
Sensors,Brand
Placem
entand
wearability
Target
shou
lder
parameters,
Perfo
rmance
Goldstandard
Task
executed
Participants
Aim
2018,[62]
Full-Text
Shim
mer
UA,FA,hand(on
centresof
mass)
Strap
FLX,
hand
SU)
7M
31.14±
5.67
17F
28±3.97
MS-P(n=
34)
13M
36.46±
13.07
21F
42.19±
12.55
iden
tifyqu
antitativeparametersfor
evaluatio
nof
ULdisabilityandrelate
itto
clinicalscores
Jung
2018,[63]
Con
ference
M-IM
U(n=5),
Xsen
sMTw
Awinda
Bilateral:
UAandFA
,trunk
Velcro
straps
ShRO
M(HTjoin
angles),
RMSE
=0.32
for
theestim
ationof
movem
entsqu
alities
usingdata
ofthe
entiredu
ratio
nof
movem
ents
Qualityof
movem
ents’label
provided
bythe
therapist
Reaching
exercise
SP(n=5),F
66.6±15.9
Y
Evaluate
movem
entsqu
ality
regarding
compe
nsationandinter-joint
coordinatio
n;exploitin
gasupe
rvised
machine
learning
approach,validate
thehypo
thesisthat
therapists’evaluation
canbe
madeconsideringon
lythe
beginn
ingmovem
entdata
Lin2018,[64]
Full-Text
IMU(n=2)
Unilateral:
UA(distal,ne
arelbo
w),FA
(dorsal
aspe
ctof
thewrist)
Strap
ShRO
M–
ShFLX-EXT
ShAB
ShER
ElbFLX
FAPR-SU
SP(n=18):
n=9,
control
grou
p(62.6±7.1)
n=9,
device
grou
p(52.2±10.2
Y)
Evaluate
thefeasibility
andefficacyof
anIMU-based
system
forup
perlim
brehabilitationin
stroke
patientsand
compare
theinterven
tioneffectswith
thosein
acontrolg
roup
Repn
ik2018,[7]
Full-Text
M-IM
U(n=7),
Myo
armband
,n=2_
with
n=8EM
Gs
built-in
,Thalamic
labs
Bilateral:
-M-IM
U:Backof
the
hand
,wrist,UA
(distal,ne
arelbo
w),
sternu
m-M
YO:FA(in
the
proxim
ityof
elbo
wjoint)
Straps,arm
band
ShRO
M(HTjointangles)
–ARA
Ttasks:19
movem
ents
divide
din
4subtests(grasp,
grip,p
inch,g
ross
arm
movem
ent)
SP(n=28)
18M,10F
57±9.1Y
HS(n=12)
9M,3
F36
±8Y
Quantify
ULandtrun
kmovem
ent
instroke
patients
accaccelerometer,g
yrgy
roscop
e,mag
nmag
netometer,O
LEOptical
Line
arEn
code
r,IMUInertia
lMeasuremen
tUnit,M-IM
UMag
neto
andInertia
lMeasuremen
tUnit,UAUpp
erArm
,FAFo
rearm,R
OM
Rang
eof
motion,
HThu
merotho
racic,GHglen
ohum
eral,Shshou
lder,w
riwrist,elbelbo
w,FLX-EXT
flexion
-exten
sion
,PR-SU
pron
ation-supina
tion,
AB-ADab
duction-ad
duction,
IERinternal-externa
lrotation,
RMSE
root
meansqua
reerror,
r=correlation,
ICCIntraclass
Correlatio
nCoe
fficients,go
ngo
niom
eter,H
SHealth
ysubject,CG
controlg
roup
,SPStroke
Patie
nt,M
S-PMultip
leSclerosisPa
tient,P
patie
nt,M
male,
Ffemale,
YYe
arsold,
FASFu
nctio
nal
AbilityScale,
ARA
TActionresearch
arm
test
Carnevale et al. BMC Musculoskeletal Disorders (2019) 20:546 Page 10 of 24
Table
3Shou
lder
motionmon
itorin
gforapplicationin
patientsun
dergoing
rehabilitation
Reference,
Year,Type
of publication
Sensors,Brand
Placem
entandwearability
Target
shou
lder
parameters,
Perfo
rmance
Gold
standard
Task
executed
Participants
Aim
Cutti2008,
[37]
Full-Text
M-IM
U(n=4),
Xsen
sMT9B
Unilateral:
Thorax
(flat
portionof
the
sternu
m),Scapula(cen
tral
third
,aligne
dwith
the
cranialedg
eof
thescapular
spine),hum
erus
(cen
tral
third
,slightlypo
sterior),
FA(distal)
Dou
ble-side
dtape
,elastic
cuff
ShRO
M(HTandST
joint
angles)
RMSE
=0.2°-3.2°
VICO
NExp1:elbFLX-EXT,PR-
SU shFLX-EXT,IER,sh
EL-DE
andP-R
Exp2:tasksin
Exp1
+sh
IER(arm
abdu
cted
90°),
shAB-AD(fron
talp
lane
),HTN
(sagittalandfro
ntal
plane)
HS(n=1,M)
Develop
aprotocol
tomeasure
humerotho
racic,scapulotho
racicandelbo
wkine
matics
Parel2012,
[65]
Full-Text
M-IM
U(n=3),
Xsen
sMTx
Unilateral:
thorax,scapu
la,hum
erus
Elastic
cuff,adhe
sive
ShRO
M(HTandST
joint
angles)
–Hum
erus
FLX-EXT(sagit-
talp
lane
)andAB-AD
(scapu
larplane)
HS(n=20)
7F,13M
28.3±5.5Y
P(n=20)
8F,12M
43.9±19.9Y
Assesstheinta-andinter-op
erator
agreem
ent
ofISEO
protocol
inmeasurin
gscapuloh
umeral
rhythm
Dapon
te2013a,[66]
Con
ference
M-IM
U(n=2),Zolertia
Z1Unilateral:
UA,FA
Brace
ShRO
M–
ShAB-AD
ElbFLX
HS(n=1)
Discuss
design
andim
plem
entatio
nof
aho
merehabilitationsystem
Dapon
te2013b,
[67]
Con
ference
M-IM
U(n=2),Zolertia
Z1Unilateral:
UA,FA
Brace
ShRO
M,
Test1:
max
gap=
6.5°
(roll)
5.2°
(pitch)
11.6°(yaw
)
Test1:
Tecno
Body
MJS
Test2:BTS
Test1:shou
lder
IR,EL-
DEandho
rizon
talFLX-
EXT
Test2:elbo
wFLX-EXT
(along
sagittalandho
ri-zontalplane)
HS(n=1,M)
Validationof
aho
merehabilitationsystem
Pan2013,
[68]
Full-Text
Acc
(n=2),
LIS3LV02DQ
Acc
built-in
aSm
artpho
ne(n=
1)
Unilateral:
-acc:U
A,tho
rax
-Smartpho
ne:w
rist
Strap,
Arm
band
ShRO
M(HTjointangles)
–Touchear
Use
finge
rsto
clim
bwall
Pend
ulum
clockw
ise
andcoun
terclockw
ise
Active-assisted
stretch
fore
andside
,raises
hand
from
back
HS(n=10)
3M,7
F20-25Y
P(n=14)
5M,9
F44–67Y
Describede
sign
andim
plem
entatio
nof
ashou
lder
jointho
me-basedrehabilitation
mon
itorin
gsystem
Thiemjarus
2013,[69]
Con
ference
Acc
(n=1),
AnalogDevice
(acc:A
DXL330)
Magn(n=1),
Hon
eywell
(magn:HMC5
843)
Unilateral:
UA(proximal)o
rwrist,l
eftor
right
Strap
ShRO
M,
RMSE
=0.86°-
5.05°
–Sh
FLX-EXT,AB-AD,hori-
zontalAB-AD,IER
HS,(n=23)
20–55Y
Evaluate
theeffect
ofsensor
placem
enton
theestim
ationaccuracy
ofshou
lder
ROM
Carnevale et al. BMC Musculoskeletal Disorders (2019) 20:546 Page 11 of 24
Table
3Shou
lder
motionmon
itorin
gforapplicationin
patientsun
dergoing
rehabilitation(Con
tinued)
Reference,
Year,Type
of publication
Sensors,Brand
Placem
entandwearability
Target
shou
lder
parameters,
Perfo
rmance
Gold
standard
Task
executed
Participants
Aim
Rawashd
eh2015,[70]
Con
ference
M-IM
U(n=1),
InvenSen
se(gyr:ITG
-3200)
AnalogDevice(acc:A
DXL
345)
Hon
eywell(magn:HMC5883L)
Unilateral:
UA(lateral)
Strap
ShRO
M–
7sh
rehabilitation
exercises
2sportsactivities
HS(n=11)
Describeade
tectionandclassification
metho
dof
shou
lder
motionge
stures
that
can
beused
topreven
tshou
lder
injury
Álvarez
2016,[71]
Full-Text
M-IM
U(n=4),
Xsen
sMTx
Unilateral:
Back
ofthehand
,FA
(nearwrist),U
A(near
elbo
w),back
Wristband,
Velcro
strap,
elastic
band
ShRO
M,
Labtest:m
ean
error=
0.06°
(FLX)
1.05°(lateral
deviation)
Labtest:
robo
ticwrist
Test1:Mou
ntingof
ashockdu
mpe
rsystem
Test2:ho
ldingatablet
forlong
perio
dsTest3:elbo
wFLX-EXT
Test1:
Mechanical
worker(n=1)
Test2:worker
ofa
commercial
centre
(n=1)
Test3:patient
(n=1)
Dem
onstrate
thefeasibility
ofan
IMU-based
system
tomeasure
uppe
rlim
bjointangles
inoccupatio
nalh
ealth
Lee2016,
[72]
Con
ference
Strain
sensor
(n=2),
MWCNT,Hyosung
:multi-walled
carbon
nano
tube
s,Eco-
Flex0030,Smoo
th-On:silicon
rubb
er
Unilateral:Shou
lder
Skin
adhe
sive
ShRO
M,RMSE<
10°
OptiTrack
ShFLX-EXT
ShAB-AD
HS(n=1)
Validatesensorsandcalibratio
nmetho
destim
atingtw
oshou
lder
jointangles
Tran
eVajerano
2016,[73]
Con
ference
M-IM
U(n=2),
Shim
mer2r
Unilateral:
UA(distal,ne
arelbo
w),
FA(distal,ne
arwrist)
Straps
ShRO
M(HTjointangles)
–Perio
dicarm
movem
ents
HS(n=1)
Validatean
algo
rithm
topred
ictthereceived
sign
alstreng
thindicator(RSSI)andthefuture
joint-anglevalues
oftheuser
Rawashd
eh2016,[74]
Full-Text
M-IM
U(n=1),
InvenSen
se(gyr:ITG
-3200)
AnalogDevice(acc:A
DXL
345)
Hon
eywell(magn:HMC5883L)
Unilateral:
UA(cen
tralthird
)Straps
ShRO
M(HTmotion)
Visual
observation
7sh
rehabilitation
exercises,baseball
throws,volleyserves
HS(n=11)
25±7Y
Validateade
tectionandclassification
algo
rithm
ofup
perlim
bmovem
ents
Wu2016,
[75]
Full-Text
M-IM
U(n=3),Bluetoo
th3-
SpaceSensor,YEI
Unilateral:FA
(nearwrist),
UA(nearelbo
w),thorax
(shifted
totherig
ht)
Strap,
band
age
ShRO
M(HTjointangles)
–12
gestures
common
indaily
life:3staticand9
dynamic
HS(n=10)
9M,1
F22–24Y
Evaluate
accuracy,recalland
precisionof
age
sturerecogn
ition
system
Burns2018,
[27]
Full-Text
Acc
andgyrbu
ilt-in
asm
art-
watch
(n=1),A
ppleWatch
(Series2&3
)
Unilateral:
Wrist
Wristband
ShRO
M–
Pend
ulum
AB
Forw
ardEL
IR ER TrapeziusEXT
Upright
row
HS(n=20)
14M,6
F19–56Y
Evaluate
perfo
rmance
ofacommercial
smartw
atch
tope
rform
homeshou
lder
physiotherapymon
itorin
g
Carnevale et al. BMC Musculoskeletal Disorders (2019) 20:546 Page 12 of 24
Table
3Shou
lder
motionmon
itorin
gforapplicationin
patientsun
dergoing
rehabilitation(Con
tinued)
Reference,
Year,Type
of publication
Sensors,Brand
Placem
entandwearability
Target
shou
lder
parameters,
Perfo
rmance
Gold
standard
Task
executed
Participants
Aim
Esfahani
eNussbaum
2018,[11]
Full-Text
Textile
sensors(prin
ted)
n=11
Bilateral:
Shou
lder
(n=6),low
back
(n=5)
Und
ershirt
ShRO
M,m
ean
error=
9.6°
for
shangle
estim
ation
M-IM
U(Xsens
MTw
Awinda)
ShAB
ShFLX-EXT
ShIER
Left/right
side
bend
ing
Trun
kFLX-EXT
Trun
krotleft/right
HS(n=16),
10M,6
F21.9±3.3
Describeasm
artun
dershirtandevaluate
itsaccuracy
intask
classificationandplanar
angle
measuresin
theshou
lder
jointsandlow
back
Ramkumar
2018,[76]
Full-Text
Acc,gyr
andmagnbu
ilt-in
asm
artpho
ne,A
ppleiPho
neUnilateral:
UA,FA
Arm
band
ShRO
M<5°
gon
AB(coron
alplane)
forw
ardFLX(sagittal
plane)
IER(elbow
fixed
tothe
body
flexedto
90°)
HS(n=10)
5M,5
FMean27
Y
Validateamotion-basedmachine
learning
softwarede
velopm
entkitforshou
lder
ROM
accaccelerometer,g
yrgy
roscop
e,mag
nmag
netometer,IMUInertia
lMeasuremen
tUnit,M-IM
UMag
neto
andInertia
lMeasuremen
tUnit,UAUpp
erArm
,FAFo
rearm,R
OM
Rang
eof
motion,
HThu
merotho
racic,ST
scap
ulotho
racic,Sh
shou
lder,elb
elbo
w,FLX-EXT
flexion
-exten
sion
,AB-ADab
duction-ad
duction,
IERinternal-externa
lrotation,
P-Rprotraction-retractio
n,R
MSE
root
meansqua
reerror,HSHealth
ysubject,Ppa
tient,M
male,
Ffemale,
YYe
arsold
Carnevale et al. BMC Musculoskeletal Disorders (2019) 20:546 Page 13 of 24
Table
4Stud
iesfocusedon
validation/de
velopm
entof
system
s/algo
rithm
sformon
itorin
gshou
lder
motion
Reference,
Year,Typeof
publication
Sensors,Brand
Placem
entandwearability
Target
shou
lder
parameters,
Perfo
rmance
Gold
standard
Task
executed
Participants
Aim
Jung
2010,
[77]
Con
ference
IMU(n=6),
ADXL
345(acc)
LPY5150
AL(gyr)
HMC5843
(magn)
Bilateral:
UAandFA
(distalthird),thorax
andpe
lvis
Strap
Shorientation
andpo
sitio
nHDRT
system
Arm
sabovehe
adBend
arms
Bend
waist
HS(n=1)
Validatethemotiontracking
algo
rithm
El-Goh
ary
2011,[78]
Con
ference
IMU(n=2),
APD
MOpal
Unilateral:
FA(nearwrist),U
A(distalthird)
Strap
ShRO
M,
r=0.91–0.97
Eagle
Analog
System
ShFLX-EXT
ShAB-AD
ElbFLX-EXT
ElbPR-SU
HS(n=1)
Validatedata
fusion
algo
rithm
Zhang2011,
[79]
Full-Text
M-IM
U(n=3),
Xsen
sMTx
Unilateral:
UA(laterally,abo
vetheelbo
w),
FA(lateraland
flatside
oftheFA
near
thewrist),sternum
Strap,
clothing
ShRO
M,
RMSE
=2.4°
(shFLX-EXT)
RMSE
=0.9°
(shAB-AD)
RMSE
=2.9°
(shIER)
BTSSM
ART-
DFree
movem
ents
HS(n=4)
Validatesensor
fusion
algo
rithm
El-Goh
ary
2012,[80]
Full-Text
IMU(n=2),
APD
MOpal
Unilateral:
UA(m
iddlethird
,slightly
posterior),
FA(distal,ne
arwrist)
Strapband
ShRO
M,
RMSE
=5.5°
(shFLX-EXT)
RMSE
=4.4°
(shAB-AD)
VICON
ShFLX-EXT
ShAB-AD
ElbFLX-EXT
FAPR-SU
Touching
nose
Reaching
forado
orknob
HS(n=8)
Validatedata
fusion
algo
rithm
LeeeLow
2012,[81]
Full-Text
Acc
(n=2),Freescale
MMA7361
LUnilateral:
UA(nearelbo
w),FA
(nearwrist)
-
ShRO
M,
RMSE
=2.12°
(shFLX-EXT)
RMSE
=3.68°
(shrotatio
n)
IMU
(Xsens
MTx)
UAFLX-EXTandmed
ial/lat-
eralrotatio
nFA
FLX-EXT
andPR-SU
(sagittalplane)
HS(n=1)
Validatethefeasibility
oftheprop
osed
algo
rithm
Hsu
2013,
[82]
Con
ference
M-IM
U(n=2),
LSM303D
LH(acc,m
agn)
L3G4200D(gyr)
Unilateral:
UA,FA
Velcro
strap
ShRO
M,
RMSE
=1.34°-5.08°
Xsen
sMTw
,(n=2)
ShFLX,
AB,EXT,ER
andIR
HS(n=10)
8M,2
F23.3±1.33
Y
Validatedata
fusion
algo
rithm
Lambrecht
eKirsch
2014,
[16]
Full-Text
M-IM
U(n=4),
InvenSen
seMPU
-9150
chip
Unilateral:
Sternu
m,U
A,FAandhand
-
ShRO
M,
RMSE
=4.9°
(shazim
uth)
RMSE
=1.2°
(shelevation)
RMSE
=2.9°
(shIR)
Optotrack
Reaching
movem
ents
HS(n=1)
Validatesensors’accuracy
anddata
fusion
algo
rithm
Ricci2014,
[83]
Full-Text
M-IM
U(n=5),
APD
MOpal
Bilateral:
Thorax,U
A(latero-distally)
andFA
(nearwrist)
Velcro
strap
ShRO
M(HTjointangles)
–UAFLX-EXT
UAAB-AD
FAPR-SU
FAFLX-EXTThorax
rotatio
nThorax
FLX-EXT
Children
(n=40)
6.9±0.65
Y
Develop
acalibratio
nprotocol
for
Thorax
andup
perlim
bmotion
capture
Roldan-
Jimen
ezInertialsen
sorsbu
ilt-in
aSm
artpho
ne(n=1),
Unilateral:
UA
ShRO
M–
shAB,EXTwith
wristin
neutralp
osition
andelb
HS(n=10)
7M,3
FStud
yhu
merus
kine
maticsthroug
hsixph
ysicalprop
ertiesthat
correspo
nd
Carnevale et al. BMC Musculoskeletal Disorders (2019) 20:546 Page 14 of 24
Table
4Stud
iesfocusedon
validation/de
velopm
entof
system
s/algo
rithm
sformon
itorin
gshou
lder
motion(Con
tinued)
Reference,
Year,Typeof
publication
Sensors,Brand
Placem
entandwearability
Target
shou
lder
parameters,
Perfo
rmance
Gold
standard
Task
executed
Participants
Aim
2015,[84]
Full-Text
LGElectron
icsINC,
iPho
ne4
Neo
pren
earm
belt
extend
ed24.2±4.04
Yto
angu
larmob
ility
andacceleratio
nin
thethreeaxes
ofspace
Fantozzi
2016,[17]
Full-Text
M-IM
U(n=7),
APD
MOpal
Bilateral:
Sternu
m,U
A,FA,b
ack
ofthehand
Velcro
strap
ShRO
M(HTjointangles),
RMSE
<10°(sh
FLX-EXT,AB-AD,
IER)
BTSSM
ART-
DX
Simulated
front-crawland
breaststroke
swim
ming
HS(n=8),
M 26.1±3.4Y
Validateaprotocol
toassess
the3D
jointkine
maticsof
theup
perlim
bdu
ringsw
imming
Men
g2016,
[85]
Full-Text
M-IM
U(n=2),
Shim
mer2r
Unilateral:
UA(distal,ne
arelbo
w),
FA(distal,ne
arwrist)
Straps
ShRO
M,
Test2:RM
SE=
from
2.20°to
0.87°
VICON
ShFLX-EXT
ShAB-AD
ShIER
ElbFLX-EXT
ElbPR-SU
Test1:
HS(n=15),
M,19–23
YTest2:
HS(n=5)
Validatean
algo
rithm
toim
prove
accuracy
onmeasuremen
tsof
arm
jointangles
consideringtheprop
erties
ofhu
man
tissue
Crabo
lu2017,[86]
Full-Text
M-IM
U(n=3),X
sens,
MTw
2Awinda
Unilateral:
UA,scapu
la,
Sternu
mVelcro
strap,
doub
le-sided
tape
,elasticband
GHjointcenter
MRI
acqu
isition
Cross
andstar
motions
(2jointvelocities,
2rang
eof
motions)
HS(n=5)
3M,2
F36
±4Y
Evaluate
accuracy
andprecisionof
theGHJC
estim
ation
Kim
2017,
[87]
Con
ference
MYO
armband
(n=1):
contains
8EM
Gand
1IMU,ThalamicLabs,
MYO
armband
Unilateral:
UA(nearelbo
w)
Arm
band
ShRO
M–
ElbFLX(0°,45°,90°)with
shin
neutralp
osition
,elb
FLX(0°,
45°,90°)with
shFLX90°
(sagittalplane)
HS(n=1)
Introd
ucean
algo
rithm
forup
per
arm
andforearm
motionestim
ation
usingMYO
armband
Morrow
2017,[88]
Full-Text
M-IM
U(n=6),A
PDM
Opal
Bilateral:
FAandUA(lateral),he
ad,
sternu
mVelcro
straps
ShRO
M(HTjointangles),
RMSE
=6.8°±2.7°
(shelevation)
Raptor
12DigitalR
eal
TimeMotion
Capture
System
Pegtransfer
(tomim
icminim
allyinvasive
laparoscop
y)
Surgeo
nHS(n=6)
3M,3
F45
±7Y
ValidateaM-IM
Ubasedprotocol
tomeasure
shou
lder
EL,elbow
FLX,
trun
kFLX-EXTand
neck
FLX-EXTkine
matics
Rose
2017,
[89]
Full-Text
IMU(n=6),A
PDM
Opal
Bilateral:
UA(lateral),FA
(dorsal),
sternu
m,lum
barspine
Straps
ShRO
M(HTjointangles)
–Diagn
ostic
arthroscop
ysimulation
Surgeo
nHS(n=14)
Develop
anIMU-based
system
toassess
the
perfo
rmance
oforthop
aedicreside
ntswith
different
arthroscop
icexpe
riences
Tian
2017,
[90]
Con
ference
M-IM
U(n=2),A
cc:
LIS3LV02D,
Magn:HMC5843,
Gyr:ITG
3200
Unilateral:
UA,FA
Straps
ShRO
MVICON
ShFLXandelbFLX(sagittal
plane)
HS(n=1)
Validatedata
fusion
algo
rithm
Pathirana
2018,[91]
Full-Text
M-IM
U(n=1)
Unilateral:
Wrist
Strap
ShRO
MVICON,
Kine
ctForw
ardFLX-EXT
AB-AD
BackwardFLX-EXT
Horizon
talFLX-EXT
HS(n=14)
10M,4
FValidateaccuracy
androbu
stne
ssof
data
fusion
algo
rithm
usingasing
lesensor
tomeasure
shou
lder
jointangles
accaccelerometer,g
yrgy
roscop
e,mag
nmag
netometer,IMUInertia
lMeasuremen
tUnit,M-IM
UMag
neto
andInertia
lMeasuremen
tUnit,UAUpp
erArm
,FAFo
rearm,R
OM
Rang
eof
motion,
HThu
merotho
racic,GH
glen
ohum
eral,Shshou
lder,elb
elbo
w,FLX-EXT
flexion
-exten
sion
,PR-SU
pron
ation-supina
tion,
AB-ADab
duction-ad
duction,
IERinternal-externa
lrotation,
RMSE
root
meansqua
reerror,r=
correlation,
HSHealth
ysubject,
Mmale,
Ffemale,
YYe
arsold
Carnevale et al. BMC Musculoskeletal Disorders (2019) 20:546 Page 15 of 24
Carnevale et al. BMC Musculoskeletal Disorders (2019) 20:546 Page 16 of 24
of them (n = 55) or with other sensors (n = 8), or built-ininto other devices (e.g., smartphones, smartwatch) (n =6); additional studies (n = 4) utilized strain sensors formotion analysis.
B.1 wearable systems based on inertial sensors andmagnetometersAn IMU allows estimating both translational and rota-tional movements. Such sensors comprise gyroscopesthat measure angular velocity and accelerometers thatmeasure proper acceleration, i.e. gravitational force(static) and force due to movements (dynamic) [92]. Themain limitation of the gyroscopes is the issue bias due todrift. Gyroscopes do not have an external reference, asopposed to accelerometers that use gravity vector as ref-erence; in the orientation estimation, gyroscopes sufferof drift during the integration procedures. To compen-sate such issue, these sensors are combined with magne-tometers that measure magnetic field and use the Earth’smagnetic field as reference. The main limitation of mag-netometers is the interference due to the presence offerromagnetic materials in the surrounding environment[92]. We refer to these hybrid sensors as M-IMU (mag-netic and inertial measurement unit). By integrating theinformation derived from each sensor (i.e., acceleration,angular velocity and magnetic field) through sensor-fusion algorithms, M-IMUs provide an accurate estima-tion of the 3D-position and 3D-orientation of a rigidbody. The upper limb can be modelled as a kinematicchain constituted by a series of rigid segments, i.e.,thorax, upper arm, forearm and hand, linked to eachother by joints that allow relative motion among con-secutive links [17]. In the kinematic chain, the shoulderjoint consists of three degrees of freedom (DOFs)correspondent to abduction-adduction (AB-AD),internal-external rotation (IER), and flexion-extension(FLX-EXT) [15, 54, 57, 71, 79]. Shoulder rotations canbe described using Euler angles that identify the anatom-ical DOFs with the roll-pitch-yaw angles [17, 33, 37, 88].Sensor-fusion algorithms can exploit two main ap-proaches, deterministic or stochastic. The deterministicapproach includes the complementary filter that mergesa high pass filter for gyroscope data (to avoid drift) anda low pass filter for accelerometer and magnetometerdata [64, 82, 90, 92]. The stochastic approach includesthe Kalman Filter and its more sophisticated versions [7,55, 66, 67, 78–80, 91, 92]. The Kalman filter (KF) is themost used algorithm to process M-IMU and IMU datadue to its accuracy and reliability [15, 38, 54, 75, 83, 93].Wearable systems based on IMU or M-IMU include a
variable number of sensor nodes that, properly distrib-uted on each body segment of interest, provide kine-matic parameters such as joint ROM, position,orientation, and velocity. Fifty-one out of the included
studies used exclusively IMUs (n = 15) or M-IMUs (n =36). Systems performances were analyzed in terms of theagreement between results obtained from the M-IMU orIMU-based systems and those collected by a gold stand-ard system. Several types of systems were used as goldstandard, such as ultrasound-based system (e.g., ZebrisCMS-HS [29]), diagnostic imaging (e.g., Magnetic Res-onance [86]), optical-based systems (e.g., VICON [37,53, 54, 80, 85, 90, 91], BTS Bioengineering [15, 17, 56,67, 79], Eagle Analogue System [78], Optotrack [16],Optitrack [61], CODA [45, 55]), goniometer [53, 54]. Re-sults from an inertial system were benchmarked againstan ultrasound-based reference system, showing a rootmean square error (RMSE) of 5.81° and a mean error of1.80° in the estimation of shoulder angles of FLX-EXT,AB-AD and IER evaluated in the sagittal, frontal andtransversal planes, respectively [29]. Accuracy of a proto-col based on commercial inertial sensors (MT9B, Xsens)was tested and compared to a VICON system to meas-ure humerothoracic, scapulothoracic joint angles andelbow kinematics [37]. Results demonstrated high accur-acy in the estimation of upper limb kinematics with anRMSE lower than 3.2° for 97% of data pairs. A BTS ref-erence system was used to validate accuracy of a wear-able system comprised of commercial sensors (Xsens)and results showed a mean error difference of 13.82° forFLX-EXT, 7.44° for AB-AD, 28.88° for IR [15]. In aprotocol-validation study, commercial Opal sensors werecompared to a BTS system to assess upper limb jointkinematics during simulated swimming movements.Data showed a median RMSE always better than 10°considering movements of AB-AD, IER and FLX-EXT infront-crawl and breaststroke [17]. Opal wearable sensorswere compared to optical motion capture systems to es-timate shoulder and elbow angles [78, 80]. Planar shoul-der FLX-EXT and AB-AD were performed showing anRMSE of 5.5° and 4.4°, respectively [80]; a good correl-ation between the measurements performed on shouldermotion with the two systems was also found in [78] (nodata regarding measurements error were proposed).Some studies (n = 11) compared data obtained from
wearable sensors, custom or commercial, with a goldstandard to validate their own sensors data fusion algo-rithm (for more details see Table 4). Two different algo-rithms were compared to a customized KF [79].Comparing the results derived from the BTS system andthe inertial-based system (Xsens), the proposed algo-rithm showed a smaller error than the other twomethods for computing shoulder FLX-EXT (RMSE =2.4°), AB-AD (RMSE = 0.9°), IER (RMSE = 2.9°) [79]. Theaddition of the magnetometer-based heading correctionin the sensor data fusion algorithm was investigated totest the accuracy of an inertial-based motion trackingsystem using the Optotrak Certus (Northern Digital Inc.,
Carnevale et al. BMC Musculoskeletal Disorders (2019) 20:546 Page 17 of 24
Waterloo, ON, Canada) as reference. Results showed aRMSE of 4.9°, 1.2° and 2.9° for shoulder azimuth, eleva-tion and internal rotation, respectively [16].Four studies used only accelerometers [42, 47, 49, 81].
Systems performance analysis in measurement of armmotion, showed a RMSE lower than 3.5° and 3.68° forshoulder ROM when results from the accelerometers-based systems were benchmarked against a goniometerand commercial M-IMUs, respectively [47, 81]. Evalu-ation of upper limbs’ physical activity was performed re-cording data of accelerometers built-in wearable deviceas ActiGraph (Pensacola, Florida, Model GT3XP-BTLE)to obtain objective outcomes in patients after reverseshoulder arthroplasty [41].Shoulder ROM has been also estimated by means of a
single sensor node which integrated an accelerometerand a magnetometer [69]. Sensor fusion algorithms ofaccelerometers and magnetometers data provide accur-ate orientation estimation in static or semi static condi-tion, e.g., in a rehabilitation session in which patientsperform slow movements [81]. M-IMUs comprised of a3D accelerometer, 3D gyroscope and 3D magnetometerare the most appropriate choice for motion tracking ei-ther in static that in dynamic condition.Two accelerometer-based sensors were combined with
those built-in a smartphone to realize a smart rehabilita-tion platform for shoulder home-rehabilitation [68]. Mo-bile phone or a smartwatch, with their built-in inertialsensor units, were used as mobile monitoring devices [27,76, 84]. These results give proof of the growing trend inthe application of commercial devices in clinical settingfor rehabilitation purposes. Data has been processed usingmachine learning algorithms to extract salient featuresand for gesture recognition related to shoulder motion. Inthese techniques, the main steps are the data collection,followed by segmentation process, feature extraction andclassification [27, 49]. For instance, the identification ofdifferent types of RC physiotherapy exercises has beenperformed processing data from inertial sensors built-in awrist-worn smartwatch [27]. Data from inertial sensorsbuilt-in a smartphone were benchmarked against a man-ual goniometer. Angular differences between a machinelearning-based application and goniometer measurementsresulted less than 5° for all shoulder ROM (i.e., AD, for-ward FLX, IR, ER) [76].Two studies combined accelerometer(s) with Optical
Linear Encoder (OLE) [68, 84]. An OLE-based systemacts as a goniometer providing measures of joint angles.Despite of the simplicity and low cost of the proposedsystems, differences in shoulder ROM estimation re-sulted not negligible when data collected by the wearablesystems were compared against an inertial-based motioncapture (i.e., IGS-190 [54]) and a fiber optics-based sys-tem (i.e., ShapeWrap [71]).
Three studies included EMG sensors in their assess-ment tool in combination with accelerometers [58],IMUs [59] and M-IMU [39]. EMG sensors placed on thebiceps, triceps [59] and deltoid muscles [39] provideadditional information about upper limb motor functionand shoulder assessment, evaluating muscles activity.Quantification of upper limb motion was executedthrough a wearable device, MYO armband by Thalamiclabs, that combines EMG sensors to record electrical im-pulses of the muscles [7, 87].
B.2 wearable systems based on strain sensorsFour studies used smart-textiles instrumented by strainsensors with piezoresistive properties to estimate kine-matic parameters and to perform motion analysis [8, 11,46, 56]. Such sensing elements are stretched or com-pressed during movements of the examined body seg-ments, with consequent variation of their electricalresistance [94, 95]. Using a M-IMU system as reference,accuracy evaluation of a smart-textile with printed strainsensors showed a mean error of 9.6° in planar motionsmeasurements of shoulder joint [11]. Shoulder kinemat-ics was assessed combining a strain sensor for scapularsliding detection with two M-IMUs for HT orientationmeasure [56].Piezoresistive strain sensors directly adhered to the skin
were used to estimate shoulder ROM; the comparison be-tween reference data from an optical-based system (i.e.,Optitrack) and strain sensors showed a RMSE less than10° in shoulder FLX-EXT and AB-AD estimation [72].
Sensors placement and wearabilityPlacement of the sensing technology on the body land-marks has shown a heterogeneous distribution linked tothe different nature of the employed technology and tothe purpose for which monitoring system was designed.With respect to the monitored upper limb, 53 out of the73 studies included in this review showed a unilateraldistribution of the sensing elements while the remainingstudies utilized a bilateral placement. Several configura-tions using different number of sensors and placementshave been investigated as reported in detail in each tableand Fig. 2.Regarding the wearability, we classified the systems in
terms of how the sensors were fixed to the human body:i) by adhesive patch, ii) by means of straps or embeddedwithin pocket, iii) the sensing element is physically inte-grated into the fabric. Four studies did not specify themethod of attachment, 12 studies have stuck sensors dir-ectly on human skin by means of adhesive patch, 52studies have attached sensors through straps or embed-ding them in modular clothing, and 5 studies have inte-grated sensors directly into garments. For more detailsrefers to Tables 1, 2, 3 and 4.
Fig. 2 Placement of sensing units (NOTE One study [90] is not included because the specific position of each sensor nodes is not so clear.Legend: N = number of studies, U = Unilateral, B = Bilateral)
Carnevale et al. BMC Musculoskeletal Disorders (2019) 20:546 Page 18 of 24
DiscussionThis paper summarizes the main features of wearablesystems that have been employed in clinical setting andresearch field to evaluate upper limb functional perform-ance and particularly for shoulder ROM assessment.Shoulder complex is characterized by the greatest mobil-ity among all human joints and, due to its complexity,reviewed articles evidenced heterogeneity on the moresuitable protocol for capturing joint ROM [96].
Wearable technologyAlthough 73% of the reviewed papers use commercialproducts for tracking joint angles, many of thesepersonalize the positioning of the sensors, the calibrationmethodology and the algorithms used to process the re-corded data. This customization makes strenuous a dir-ect comparison among protocols, especially if sensingunits of different nature (e.g., M-IMU vs. strain sensors)are used to measure the same kinematic parameters,leaving still open the issue of the protocols’ definitionwith general validity.About studies using inertial-based motion tracking
systems, most in this summary (88%), calibration proce-dures before data acquisition and data processing repre-sent a relevant issue about accuracy and reliability of thesystem. Typically, the M-IMUs are attached on the seg-ment of interest to estimate its orientation, so the cali-bration is necessary to relate sensors’ measurements to
movements of the tracked body segment. Sometimes themanufacturer suggests how to perform calibration, e.g.,positioning sensors on a flat surface [15, 35] to align co-ordinate system or assuming static anatomical position[65], as N-pose [79], to compute orientation differencesbetween segments and sensors coordinates in order toobtain sensor-to-segment alignment [56]. Dynamic orfunctional anatomical calibration has also been per-formed in some studies, but the sequence of movementsexecuted varied among these [17, 33, 55, 83]. One inter-esting improvement that may be done to have a positiveimpact on the accuracy of inertial-based motion trackingsystems, is to define a standard set of movements for theinitial calibration and a standard method of data pro-cessing by which extrapolate kinematic parameters ofhigh clinical relevance.Some works have reported remarkable results in hu-
man motion tracking using e-textile sensors [8, 46].Technological improvements in the development of con-ductive elastomers allowed to integrate such strain sen-sors directly into garments making them comfortableand unobtrusive [11, 56]. Although conductive elasto-mers ensure flexibility and performances comparablewith those of the M-IMU sensors, the main limitationsare the hysteresis, uniaxial measurements and non-negligible transient time [56]. Wearable systems basedon strain sensors are a promising technology for kine-matics analysis that may overcome the main M-IMUs
Carnevale et al. BMC Musculoskeletal Disorders (2019) 20:546 Page 19 of 24
drawbacks, as interferences due to surrounding ferro-magnetic materials, gyroscopes’ error drift and long-term use. On the other hand, errors may occur withstrain sensors-based systems in the estimation of shoul-der kinematics for their inherent hysteresis behaviour.Among wearable systems reviewed in this summary,
differences resulted in terms of sensors typology, num-ber and size, placement, and wearability features. Sen-sors placement and method of attachment must becarefully investigated as they could influence the out-comes reliability (e.g., effects of soft tissues’ artefacts).Human skeleton is covered by skin tissue and muscles.The combination of skin’s elasticity and muscle activitymay cause negative effects in the measurement of thebones’ movement. In studies where M-IMU sensorswere used to track shoulder kinematics, soft tissue prop-erties were opportunely included in mathematicalmodels to reduce soft tissue artifacts [79, 85]. The bodyfat percentage was found the main influencing factorthat negatively affects the inertial sensors’ orientation[85]. To reduce such source of error, either when sen-sors are directly adherent to the skin that embedded in atextile, sensing units should be placed as near as possibleto the bone segment to reduce soft tissue artifacts [97,98]. Wearability is a key factor to consider because itcan influence the level of patients’ acceptance [26].Thereare several relevant requirements that wearable systemsmust meet to encourage their applications in continuousmonitoring of patient status. Indeed, execution of move-ments, either in home environments or in clinical set-tings, should not be hindered by measurements systemsso they must be non-invasive, modular, lightweight, un-obtrusive and include a minimal number of sensors [33,40, 51, 56, 66, 67, 91]. Most studies have employed mag-neto and inertial-based tracking systems in which sen-sors were attached to the upper limb through Velcrostraps or including them in modular brace and garments[26, 45, 67, 82, 88].Upper limb includes the shoulder, elbow and wrist
joints (Fig. 3a). Humerus, scapula, clavicle, and thoraxconstitute the shoulder complex: humeral head articu-lates in the glenoid fossa of the scapula to form GHjoint, the AC joint is the articulation between the lateralend of the clavicle and the acromion process, the SCjoint articulates the medial end of the clavicle and thesternum and the functional ST joint allows rotationaland translational movements of the scapula with respectto the thorax [96] (Fig. 3b). The ST joint and GH jointact togheter in arm elevation according to scapular-humeral rhythm described in [99]. From a biomechan-ical point of view, shoulder complexity is justified by thehigh degree of coupling and coordination betweenshoulder joints (i.e., shoulder rhythm) and the action ofmore than one muscles over more than one joints in the
execution of a movement. Data extraction of shoulderkinematics is frequently based on movements pattern inthe sagittal, frontal and transversal planes, so monitoringof complex movements (e.g., daily activities) in multipleplanes, performed through wearable sensors, requires amore stringent evaluation and accurate interpretation.As resulted in the review, the shoulder is generally ap-proximated as a ball-and-socket joint [56]. This assump-tion provides an approximate representation of thewhole shoulder girdle (e.g., it neglects the contributionof scapular movements). A standardized protocol hasbeen proposed (i.e., The ISEO®, INAIL Shoulder andElbow Outpatient protocol) to improve the performanceof M-IMUs in the estimation of scapular kinematics, bylocating inertial sensors on the back in correspondenceof scapula [33, 35–38, 40, 65]. An adequate investigationof scapular motions may be beneficial to assess shoulderdisorders [100].For long-term monitoring of shoulder kinematics con-
sidering also scapular motions, the combination of M-IMUs and smart-textile with embedded strain sensors isa perfect balancing of accuracy, flexibility and wearability(i.e., strain sensors positioned on the scapula could in-crease the portability and acceptance of the wearablesystem for long-term monitoring of ADLs) [86].
Applicability in clinical setting and rehabilitationAlterations in the complex shoulder kinematics can de-rive by both neurological or musculoskeletal disordersand result in pain and limited movements [68]. Com-pensatory movements in patients with shoulder disor-ders are the most common consequential responses topain or to difficulty in performing free-pain movements.In such situations, information retrieved by posturemonitoring may be beneficial in clinical application andrehabilitation [26]. In the last years, the application ofwearable devices for gathering motion data outside thelaboratory settings is growing. Avoiding complex labora-tory set-up, wearable systems employed to assess upperlimb kinematics have proven to be a well-founded alter-native to obtain quantitative motions parameters. Quan-titative outcomes about shoulder motions recorded bywearable sensors are beneficial in clinical practice interms of time-saving and they are becoming a promisingalternative to improve assessment accuracy overcomingthe subjectivity of clinical scales. The automatic assess-ment of motor abilities can also provide therapists a tan-gible and, therefore, measurable awareness of theeffectiveness of the treatment and the recovery pathchosen.In clinical practice, the severity level of patients’ condi-
tion with musculoskeletal disease is usually assessedthrough questionnaire-based scores [36, 42]. Algorithmsfor kinematic scores computing were developed to
Fig. 3 a Anatomy of the Upper limb; b Anatomy of the Shoulder complex
Carnevale et al. BMC Musculoskeletal Disorders (2019) 20:546 Page 20 of 24
evaluate shoulder functional performance after surgeryin subjects with GH osteoarthritis and RC diseases, elab-orating data obtained from IMU sensors [29, 31]. Highcorrelation (0.61–0.8) between shoulder kinematicscores (i.e., power score, range of angular velocity scoreand moment score) and clinical scales (e.g., DASH, SST,VAS) was found [31]. Unlike clinical scores, kinematicscores showed greater sensitivity in detecting significantfunctional changes in shoulder activity at each post-operative follow-up with respect to the baseline status[29, 31]. In a five-year follow-up study, asymmetry inshoulder movements was evaluated in patients with sub-acromial impingements syndrome. Asymmetry scores,derived from an IMU-based system, showed post-treatment improvements with greater sensitivity thanclinical scores and only a weak correlation was foundwith DASH (r = 0.39) and SST (r = 0.32) [32]. Quantita-tive evaluation of arm usage and quality of movementsin every kind of shoulder impairment contributes to out-line a clinical picture about the functional recovery andthe effectiveness of the treatment [30, 49]. Using thesame number of IMU (n = 3) and the same placementon both humeri and sternum, the shoulder function wasevaluated before and after treatment, in patients under-went surgery for RC tear [5, 34]. Results showed signifi-cative differences in movements frequency betweenpatients and control group during activities of daily life[5], with limited use of arm at 3months after surgery[34]. With a bilateral configuration based on 5 IMU,shoulder motion was assessed to extrapolate relevantclinical outcomes about Total Shoulder Arthroplasty(TSA) and Reverse Total Shoulder Arthroplasty (RTSA)[6]. Patients underwent either TSA or RTSA showedshoulder ROM below 80° of elevation, indiscriminately;
but, on average, patients treated with RTSA performedmovements above 100° less frequently [6]. Objectivemeasurements (i.e., mean activity value and activity fre-quency) of limb function after RTSA did not show sig-nificant improvements 1 year after surgery, despiteDASH scores and pain perception have improved com-pared to preoperative outcomes [41].In patient with neurological impairments (e.g., stroke), as-
sessments of motor abilities performed through wearablesens
Recommended