Animals 2015,5, 861-885; doi:10.3390/ani5030388 OPEN ACCESS
animals
ISSN 2076-2615
www.mdpi.com/jou nal/animals
Re iew
Lameness De ec ion in Dai y Cows: Pa 2. Use o Senso s o
Au oma ically Regis e Changes in Locomo ion o Beha io
Annelies Van Nu el 1,*, Ing id Zwe aeghe 1, S ephanie Van Weyenbe g 1, Ma i Pas ell 2,
Vi i M. Tho up 3,4, Claudia Bah 5, Ba Sonck 6,7 and Wou e Saeys 8
1Technology and Food Science Uni , P ecision Li es ock Fa ming, Ins i u e o Ag icul u al and
Fishe ies Resea ch (ILVO), Bu gemees e an Gansbe ghelaan 115 bus 1, 9820 Me elbeke, Belgium;
E-Mails: ing id.zwe aeghe @il o. laande en.be (I.Z);
s ephanie. anweyenbe g@il o. laande en.be (S.V.W.)
2Na u al Resou ces Ins i u e Finland (Luke), G een Technology, Koe ilan ie 5, 00790 Helsinki,
Finland; E-Mail: [email p o ec ed]
3INRA, UMR 791 Sys emic Modelling o Ruminan Nu i ion, 16 ue Claude Be na d, 75231 Pa is
cedex 05, F ance
4Ag oPa isTech, UMR 791 Sys emic Modelling o Ruminan Nu i ion, 16 ue Claude Be na d,
75231 Pa is cedex 05, F ance; E-Mail: i[email p o ec ed]
5Di ision Measu e, Model and Manage BioResponses, Depa men o Biosys ems, Ka holieke
Uni e si ei Leu en, Kas eelpa k A enbe g 30 bus 2456, 3001 He e lee, Belgium;
E-Mail: Claudia.bah @biw.kuleu en.be
6Animal Sciences Uni , Ins i u e o Ag icul u al and Fishe ies Resea ch (ILVO), Scheldeweg 68,
9090 Melle, Belgium; E-Mail: ba .sonck@il o. laande en.be
7Depa men o Biosys ems Enginee ing, Facul y o Bioscience Enginee ing, Ghen Uni e si y,
Coupu e links 653, 9000 Gen , Belgium
8Di ision Mecha onics, Bios a is ics and Senso s (MeBioS), Depa men o Biosys ems, Ka holieke
Uni e si ei Leu en, Kas eelpa k A enbe g 30 bus 2456, 3001 He e lee, Belgium;
E-Mail: wou e [email protected] en.be
*Au ho o whom co espondence should be add essed;
E-Mail: annelies. annu [email p o ec ed].
Academic Edi o : Jon Huxley
Recei ed: 29 May 2015 / Accep ed: 23 July 2015 / Published: 28 Augus 2015
Simple Summa y: As lame cows p oduce less milk and end o ha e o he heal h p oblems,
inding and ea ing lame cows is e y impo an o a me s. Senso s ha measu e
Animals 2015,5862
beha io s associa ed wi h lameness in cowscan help by ale ing he a me o hose cows
in need o ea men . This e iewgi es an o e iew o senso s o au oma ed lameness
de ec ion and discussessome p ac ical conside a ions o in es iga ing and applying such
sys ems inp ac ice.
Abs ac : Despi e he esea ch on oppo uni ies oau oma ically measu e lameness in
ca le, lameness de ec ion sys ems a e no widely a ailable comme cially and a e only used
on a ew dai y a ms. Howe e , a me s need o be awa e o he lame cows in hei
he ds in o de ea hemp ope ly and in a imely ashion. Many pape s ha e ocused on
he au oma edmeasu emen o gai o beha io al cow cha ac e is ics ela ed o lameness.
Ino de o such au oma ed measu emen s o be used in a de ec ion sys em, algo i hms o
dis inguish be ween non-lame and mildly o se e ely lame cowsneed o be de eloped and
alida ed. Few s udies ha e eached his la e s ageo he de elopmen p ocess. Also,
compa ison be ween he di e en app oachesis impeded by he wide ange o p ac ical
se ings used o measu e he gai o beha io alcha ac e is ic (e.g., measu emen s du ing
no mal a ming ou ine o du ingexpe imen s; cows guided o walking a hei own speed)
and by he di e en de ini ions o lame cows. In he majo i y o he publica ions, mildly lame
cowsa e included in he non-lame cow g oup, which limi s he possibili y o alsode ec ing
ea ly lameness cases. In his e iew, s udies ha used senso echnology o measu e
changes in gai o beha io o cows ela ed o lamenessa e discussed oge he wi h p ac ical
conside a ions when conduc ing lameness esea ch. In addi ion, o he p e equisi es o any
lameness de ec ion sys em on a ms (e.g., need o ea ly de ec ion, eal- ime measu emen s)
a e discussed.
Keywo ds: Lameness; ea ly de ec ion; on- a m; dai y ca le
1. In oduc ion
To p ope ly ackle he lameness p oblem, a me s need o be awa e o he numbe o lame cows
in hei he d and he se e i y o hei lameness. The commonly accep ed me hodologies o quan i y
lameness ely on iden i ying changes in he gai and pos u e o he cows and a e discussed in he i s
pa o his e iew [1]. In p ac ice, his is done using subjec i e me hods such as isual obse a ions
leading o locomo ion sco es by he a me , an employee, a e e ina ian o an ag icul u al consul an .
Subjec i e locomo ion sco ing is quick o apply, inexpensi e and easy o pe o m. Howe e , i is ime
consuming when used o sco e he whole he d, and locomo ion sco ing is also subjec i e and equen ly
esul s in disc epancies be ween and wi hin obse e s [2,3]. Mo eo e , mos a me s unde es ima e he
p e alence o lameness in hei he d by a ac o o ou o mo e [4].
Since he 1980s, a ious senso s and echnologies ha e been in es iga ed o hei po en ial o measu e
heal h indica o s om indi idual cows [5]. Au oma ed sys ems i he end o inc easing he d sizes and
less ime o a me s o moni o he he ds, le alone he indi idual cows. In addi ion, an au oma ed
Animals 2015,5863
lameness de ec ion sys em migh e en be mo e sensi i e han adi ional obse a ional me hods o
lameness de ec ion [2].
Au oma ic measu emen sys ems could suppo he dai y a me and ackle he p oblem o he
isual sco ing sys ems, esul ing in mo e objec i e measu emen s. Howe e , he aw da a collec ed
om such senso -based echniques s ill ha e o be ansla ed in o unde s andable gai a iables ha a e
unc ionally ele an o ca le gai (and he e o e ela ed o gai a ibu es used in locomo ion sco ing
sys ems). This e iew summa izes se e al app oaches o au oma ically measu e lameness and ela ed
cha ac e is ics in dai y cows. The wide ange o p ac ical se ings du ing he expe imen s, oge he wi h
he de ini ions used o dis inguish non-lame om lame cows, impede compa ison be ween he di e en
au oma ed app oaches. The ac o s obs uc ing such compa ison and he need o speci ic cha ac e is ics
( eal- ime/au oma ed/con inuous measu emen s, le el o ea ly de ec ion, se -up size) o an au oma ed
lameness de ec ion sys em a e discussed.
2. Au oma ic Gai and Beha io Measu emen s and Lameness De ec ion
Se e al echniques ha e been p oposed o au oma ic gai analysis, such as o ce pla o ms,
elec omyog aphy, accele ome e s and image-based echnologies. In hei e iew, Ru en e al. [6]
di ided he scien i ic publica ions in o ou ca ego ies: (I) echniques ha measu e some hing abou
he cow (e.g., ac i i y); (II) in e p e a ions ha summa ize changes in he senso da a (e.g., dec ease in
ac i i y) o p oduce in o ma ion abou he cow’s s a us (e.g., lameness); (III) in eg a ion o in o ma ion
whe e senso in o ma ion is supplemen ed wi h o he in o ma ion (e.g., economic in o ma ion) o
p oduce ad ice (e.g., whe he o ea a cow o no ); and (IV) he a me making a decision o he
senso sys em making he decision au onomously (e.g., he imme o e e ina ian is con ac ed).
In 2013, 38 publica ions on lameness we e desc ibed in he e iew by Ru en e al. [6], bu only
hal o hem included da a in e p e a ion, o which only one ou o i e had alida ed hei algo i hm,
o en wi hou epo ing he sensi i i y and speci ici y. The main eason o his low ac ion o alida ed
algo i hms is ha mos s udies s ill ocused on ca ego y I: gaining in o ma ion on cow cha ac e is ics
ela ed o lameness using senso echnology. In he nex sec ion, di e en measu ing echniques used
o gain in o ma ion on speci ic lameness- ela ed cha ac e is ics in cows a e summa ized based on he
used senso echnology. Only hose pee e iewed pape s ha compa ed he senso da a wi h locomo ion
sco ing as e e ence we e conside ed. O he e e ences such as lesion sco ing we e a ely inco po a ed,
as his e iew ocuses on he de ec ion o lameness in gene al and no on he de ec ion o hoo lesions.
2.1. Load Cells
2.1.1. Measu emen o Walking Cows
Rajkondawa e al. [7] de eloped a 1D g ound eac ion o ce sys em ha measu ed g ound eac ion
o ces (GRF) when he cows walked o e 2 pa allel o ce pla o ms 1.98 m long and calcula ed 7
a iables on he le and igh hind legs o he cow. Bo h sides o he pla o m had 4 load cells unde he
loo pla e. Wi h his sys em he peak GRF was measu ed o 11 non-lame and 12 lame cows [8]. In 2006,
mo e a iables we e added o he lis o dis inguishing lame om non-lame cows: peak GRF, a e age
Animals 2015,5864
GRF, s ance ime, impulse (i.e., in eg al o GRF o e ime), and a ea unde he Fou ie ans o med cu e
o GRF [9].
Based on hese publica ions, he au oma ic lameness de ec ion sys em S epMe ix (BouMa ic,
Madison, WI, USA) was in oduced in o he ma ke in 2008. The model in his sys em used 5 limb
mo emen a iables and calcula ed he p obabili y o one o he hind limbs being lame (SMX sco e).
The S epMe ix was es ed in a ield ial by Bicalho e al. [10]. They epo ed a high speci ici y (93.8%),
bu a low sensi i i y (22.2%) using he comme cial se ings o he S epMe ix . Hence, lame cows we e
s ill being misclassi ied by he model. Liu e al. [11] imp o ed he speci ici y and o a lesse deg ee
he sensi i i y up o 43% by using ans o med limb mo emen a iables. Howe e , in ano he s udy by
Liu e al. [12] he sensi i i y o his lameness de ec ion sys em was s ill limi ed (51.9%). The au ho s
asc ibed hese poo esul s o he ailu e o ans e ing load o he limb whe e he cow showed highe
pain eac ions o whe e locomo ion abno mali ies o lesions we e sco ed.
Fo ces exe ed by he hoo es on he g ound can be measu ed using o ce pla es—usually ins alled in
he g ound— ha gi e in o ma ion in one o h ee dimensions ( e ical, o wa d/backwa d and sidewa ds
ho izon al o ces, Fz, Fy and Fx espec i ely). Use ul in o ma ion o he g ound eac ion o ce can only
be collec ed when no mo e han one limb a a ime is on he o ce pla e [13]. In ca le, using a o ce
pla e only 0.52 m long, app oxima ely 30 epe i ions we e necessa y o ob ain one measu emen whe e
he hoo was ully measu ed by he o ce pla e [14]. Howe e , using a se o pa allel o ce pla es ha
we e 2.07 m long, on a e age 3.2 alid s eps we e ob ained in jus one epe i ion [15].
Tho up e al. [16] used he 3D o ce pla e sys em desc ibed in Skjø h e al. [15] and showed ha ull
e ical o ce cu e symme y be ween le and igh leg pai s was lowe in lame cows compa ed wi h
non-lame cows. Impo an ly, hey classi ied non-lame cows as sco ing 1, whe eas lame cows sco ed 2-5,
hus using a low cu -o . Howe e , hei s udy only included a small numbe o cows, so he use ullness
o ull cu e symme y o lameness de ec ion emains o be con i med by la ge s udies.
Pas ell e al. [17] in oduced he EmFi , a ma made o elec omechanical ilm which can de ec
only dynamical o ces, in o lameness esea ch. Thei p elimina y es showed ha such sys ems ha e
po en ial o sepa a e lame om non-lame cows based on he di e ences in o ce- ime beha io : lame
cows used a lowe s ep o ce and longe s ance ime wi h he lame limb compa ed o he sound limb.
A sha pe nega i e peak o he lame limb showed ha he pain ul limb was li ed much as e han he
sound limb [17].
2.1.2. Measu emen o S anding Cows
In an ea lie s udy, Pas ell e al. [18] s a ed wi h au oma ic measu emen o he weigh dis ibu ion
o he cow in a milking obo whe e he weigh o each limb was measu ed sepa a ely using load cells.
An algo i hm calcula ed he a e age weigh , he weigh a ia ion o each limb and he numbe and
equency o kicks and s eps. P elimina y esul s on a esea ch a m ga e e idence ha obse ed limb
and oo diso de s could be de ec ed [18]. Pas ell and Kujala [19] buil a model using a p obabilis ic
neu al ne wo k. Valida ion o his model showed 96.2% co ec classi ica ion o cows as being non-lame
o lame, and 100% (Se) o he lameness cases we e iden i ied wi h a speci ici y o 57.5% (AUC = 0.86).
The model could be adjus ed o highe speci ici y a he cos o sensi i i y. O all measu emen s, only
1.1% we e alsely classi ied as being lame. Pas ell and Madsen [20] used CUSUM cha s du ing a long
Animals 2015,5865
e m ollow-up s udy using he cow’s own his o ical da a o de ec lameness. They sugges ed adjus ing
he speci ic CUSUM-cha alues in o de o ha e a speci ic le el o de ec ion a e and alse ala ms
depending on he use o he sys em (e.g., on a m o lameness de ec ion, o ea ly lameness de ec ion,
o in a e e ina y p ac ice o diagnosis). This sys em was epo ed o iden i y lame cows wi h speci ic
lesions (sole ulce s and whi e line disease) mo e quickly han isual locomo ion sco ing, while join
p oblems we e mo e easily de ec ed based on he locomo ion sco es [21]. Fo he de ec ion o mildly
lame cows, epea ed measu emen s on indi idual cows we e sugges ed o de ec changes ha occu
o e ime.
Ne eux e al. [22] used a pla o m ou side he au oma ic milking sys em con aining ou eco ding
uni s wi h wo load cells each o measu e he weigh dis ibu ion o cows while s anding on di e en
su aces ( ubbe and conc e e). An adjus ed se -up was la e used by Chapinal e al. [23,24] and
Pas ell e al. [25] o measu e lameness and hoo lesions. The asymme y in weigh bea ing be ween le
and igh limbs (Leg Weigh Ra io o LWR) was ound o be a sensi i e measu e o de ec ing se e ely
lame cows (AUC = 0.87). Pas ell e al. [25] sugges ed ha a cow may su e pain when walking, which is
no as ob ious when he cow is s anding s ill. This may be because o se e al undiagnosed causes behind
unc ional lameness, such as hickened join capsules, ne e o ligamen inju y, and muscle p oblems
no a ec ing s anding. They also ound ha s anding LWR measu emen s we e highly co ela ed wi h
locomo ion sco e o cows wi h sole ulce s (R2= 0.79), and ha some sole ulce s seemed o cause
changes in LWR while s anding bu did no esul in isible lameness du ing walking. On he o he hand,
LWR was less success ul a iden i ying cows wi h only haemo hages, and was unable o disc imina e
be ween cows wi h no hoo lesions and cows wi h digi al de ma i is.
Chapinal and Tucke [26] used a simila sys em o alida e au oma ically measu ed s eps, while cows
s and on he pla o m. Compa ed o coun ing s eps on ideo oo age, his sys em ga e a high sensi i i y
and speci ici y (>96%) in coun ing s eps.
Using a p essu e dis ibu ion pla e (RsScan), an de Tol e al. [27] measu ed he p essu e dis ibu ion
unde one hoo a a ime in s anding cows. an de Tol e al. [28] used his se -up wi h an addi ion o a
o ce pla e o measu e GRF and p essu e dis ibu ions while walking. Howe e , no es s wi h lame and
non-lame cows we e pe o med wi h his sys em.
2.2. P essu e-Sensi i e Posi ion Ma
Mae ens e al. [29] de eloped he Gai wise sys em. I is a p essu e-sensi i e posi ion ma which
p o ides spa io- empo al and ela i e o ce in o ma ion o wo comple e gai cycles o cows walking
o e he measu emen zone (Figu e 1). The Gai wise sys em au oma ically measu es he gai o he
cows a e milking ( wice a day) wi hou human in e e ence. Gai a iables a e a ailable a ew seconds
a e he cow has passed he measu emen zone. The aw da a o his senso is ans o med o 20 basic
a iables ha desc ibe he gene al gai o cows walking o e he measu emen zone oge he wi h an
addi ional 10 mo e speci ic gai a iables ha a e closely ela ed o gai cha ac e is ics used in locomo ion
sco ing sys ems: “s ide leng h”, “s ide ime”, “s ance ime”, “s ep o e lap”, “abduc ion” and a iables
co e ing asymme y be ween le and igh limbs: “asymme y in s ep wid h”, “asymme y in s ep
leng h”, “asymme y in s ep ime”, “asymme y in s ance ime” and “asymme y in ela i e o ce”. A
s udy by Van Nu el e al. [30] e ealed ha a iables o speed and asymme y measu ed by he Gai wise

Animals 2015,5866
sys em we e closely ela ed o locomo ion sco es gi en by obse e s based on ideos o cows walking
o e he Gai wise sys em and ma ched he Gai wise a iables. The i s de ec ion model based on hese
speci ic a iables could classi y 84% o he cows co ec ly, wi h a sensi i i y o 85%, 76% and 90% and
a speci ici y o 86%, 89% and 100% o he de ec ion o non-lame, mildly lame and se e ely lame cows
espec i ely [29]. In a ollow-up s udy, a iables o gai inconsis ency we e es ed using wo case-con ol
s udies, e ealing he po en ial o hese new a iables o he de ec ion o mildly lame cows [31]. Indeed,
de ec ion o he mildly lame cows in he model based on he a iables o inconsis ency imp o ed o a
sensi i i y o 88% wi h a speci ici y o 87% [32]. Mo eo e , he esul s o he case con ol s udies
sugges ed ha he p oblem o loca ing he cause o lameness migh be de i ed om he a iables ha
change signi ican ly [31]. A es - e es s udy showed ha measu emen s wi h he Gai wise sys em we e
highly epea able wi hin cows [31].
Animals 2015, 5 865
based on hese speci ic a iables could classi y 84% o he cows co ec ly, wi h a sensi i i y o 85%,
76% and 90% and a speci ici y o 86%, 89% and 100% o he de ec ion o
non-lame, mildly lame and se e ely lame cows espec i ely [29]. In a ollow-up s udy, a iables o gai
inconsis ency we e es ed using wo case-con ol s udies, e ealing he po en ial o hese new a iables
o he de ec ion o mildly lame cows [31]. Indeed, de ec ion o he mildly lame cows in he model based
on he a iables o inconsis ency imp o ed o a sensi i i y o 88% wi h a speci ici y o 87% [32].
Mo eo e , he esul s o he case con ol s udies sugges ed ha he p oblem o loca ing he cause o
lameness migh be de i ed om he a iables ha change signi ican ly [31]. A es - e es s udy showed
ha measu emen s wi h he Gai wise sys em we e highly epea able wi hin cows [31].
(A)
(B)
(C)
Figu e 1. Illus a ion o he da a acqui ed by he Gai wise sys em a e da a
condi ioning and iden i ica ion o each de ec ed imp in . (A) Loca ions o each
oo imp in on he measu emen zone in x ( ans e se) and y (longi udinal)
coo dina es. Each oo was measu ed h ee imes (Red = igh on ; black = igh
hind; yellow = le on ; g een = le hind). (B) The same da a a e ep esen ed
wi h a ime axis o show he du a ion o each imp in . (C) Colou s on he imp in
ep esen mean p essu e om each senso du ing each oo all; he ela i e scale
is om g ey (lowes ela i e p essu e) h ough cyan, yellow, magen a and ed o
blue (highes ela i e p essu e) [78].
Figu e 1. Illus a ion o he da a acqui ed by he Gai wise sys em a e da a condi ioning
and iden i ica ion o each de ec ed imp in . (A) Loca ions o each oo imp in on he
measu emen zone in x ( ans e se) and y (longi udinal) coo dina es. Each oo was
measu ed h ee imes (Red = igh on ; black = igh hind; yellow = le on ; g een = le
hind). (B) The same da a a e ep esen ed wi h a ime axis o show he du a ion o each
imp in . (C) Colou s on he imp in ep esen mean p essu e om each senso du ing each
oo all; he ela i e scale is om g ey (lowes ela i e p essu e) h ough cyan, yellow,
magen a and ed o blue (highes ela i e p essu e) [78].
Animals 2015,5867
2.3. Vision Techniques
Flowe e al. [33] we e he i s o use ision echniques wi h body ma ke s o measu e empo al and
spa ial gai cha ac e is ics in cows ela ed o lameness. Song e al. [34] used ideo images o walking
cows wi hou body ma ke s o au oma ically measu e s ep o e lap as a ele an gai cha ac e is ic o
lameness de ec ion. Pluk e al. [35] w o e an image-based algo i hm o calcula e he s ep o e lap
and compa ed he ou pu o his algo i hm o he locomo ion sco e gi en by a ained obse e . The
au oma ically measu ed s ep o e lap was ound o be highly co ela ed wi h manual locomo ion sco ing
(R2= 0.81). In an ini ial expe imen on 15 cows, he measu ed s ep o e lap seemed o be signi ican
o he dis inc ion be ween non-lame and mildly lame cows. Howe e , in a second expe imen using a
simpli ied sco ing sys em on 104 cows, his dis inc ion was only seen be ween non-lame and se e ely
lame cows o he minimal s ep o e lap and be ween mildly and se e ely lame cows o he maximal
s ep o e lap. When combining a came a sys em wi h he Gai wise sys em, an algo i hm o au oma ically
calcula e ouch and elease angles and ange o mo ion was de eloped [36]. A dec eased ange o
mo ion and inc eased elease angles o he on legs could de ec changes in locomo ion sco es in a
la ge pe cen age o he cows (76%).
Ano he p omising a iable o lameness de ec ion is he back a ch. Pou sabe i e al. [37]
au oma ically calcula ed he back a ch o each cow du ing walking by i ing a ci cle h ough h ee
selec ed poin s on he side iew con ou line o he back spine. The a e age in e se adius o ou
ames displaying he hind ee in con ac wi h he g ound ( wo ames o each hind oo ) was calcula ed
o each cow. Based on his cu a u e alue, a sco e ep esen ing he lameness s a us was gi en o
he indi idual cow. The esul s o he algo i hm we e again compa ed o locomo ion sco es esul ing
in a co ec classi ica ion o mo e han 96%. This idea was u he elabo a ed and a Body Mo emen
Pa e n was used o desc ibe he mo emen o he back and head o he cows using a side iew [38]. An
algo i hm based on his Body Mo emen Pa e n was es ed unde a m condi ions by Viazzi e al. [39],
who epo ed a co ec classi ica ion o 81% and 91% using bo h h esholds, a he popula ion le el and
a he indi idual le el, espec i ely.
The algo i hm p oposed by Viazzi e al. [39] was no ully au oma ic because i s ill equi ed wo
manual p ocedu es o selec he ames in he ideo whe e he hind hoo es bea weigh and o segmen
he back con ou line om he images. The ollow-up s udy by Van He em e al. [40] showed ha
image segmen a ion in hese 2D images was p oblema ic in eal a m condi ions. Se e al challenges
o he sys em we e summa ized in he wo k o Pou sabe i e al. [41], Van He em e al. [40] and
Viazzi e al. [42]; (1) lack o space on a ms o ins all a side- iew came a; (2) changing ligh ing
condi ions causing noise and shadows in he images ha impede ex ac ion o he back pos u e and
(3) con inuous backg ound changes ha in e e e wi h cow segmen a ion om he images. Also, he
measu emen s o Pou sabe i e al. [37] could no be au oma ed, because he ames had o be manually
examined o selec he momen in which he cow placed he hind limb on he g ound in o de o selec
he ollowing ames o be p ocessed au oma ically [42].
To o e come he di icul ies wi h back pos u e analysis based on side- iew images om 2D came as,
Van He em e al. [43] and Viazzi e al. [42] es ed he use o 3D came as in op iew. The accu acy
o he applica ion o bo h algo i hms o de ec ion o lame cows was compa able: 91% and 90% o
he 2D and 3D came a sys ems, espec i ely [42]. In he s udy by Van He em e al. [43], he co ec
Animals 2015,5868
classi ica ion a e based on he 5-poin scale, a ole an scale (whe e one uni di e ence in locomo ion
sco e be ween he model and e e ence was accep ed) and a bina y scale o isual locomo ion sco ing
(non-lame = locomo ion sco e 1 and 2; lame = locomo ion sco e 3 o 5) we e 60.2%, 90.9% and 81.2%,
espec i ely. In his s udy, only cows ha we e sco ed wi hin a one-uni sco e ange in ou consecu i e
measu emen s we e used in o de o educe he subjec i i y o he obse e and o inc ease eliabili y o
he locomo ion sco es. Viazzi e al. [42] epo ed ha using he 3D came a op iew o e comes mos o
he limi a ions o he 2D came a app oach: (1) op iew came as a e easie o ins all in he ba n because
no addi ional space a he side is needed o ge good images o he cows; (2) he dep h in o ma ion
o he 3D came a made changing backg ounds and shadows less in usi e and enhanced he au oma ic
segmen a ion o di e en cows. Limi a ions o he se -up wi h he 3D came a we e alloca ed o he high
sensi i i y o he came a o na u al ligh , small ield o iew and he lack o in o ma ion o o he gai
a iables wi h his app oach [42]. The use o 3D came as allowed ully au oma ic ideo eco ding and
p ocessing. Howe e , manual iden i ica ion o he cows was s ill needed [43].
D i en by he posi i e esul s o in a ed he mog aphy o lameness de ec ion in ho ses [44],
his echnique was es ed o de ec hoo lesions in ca le [45]. Bo h Alsaaod and Büsche [46] and
S okes e al. [47] used in a ed he mog aphy o in es iga e he possible ole o his echnique in de ec ing
he p e e ed ime o imming o ea men in e en ion in p e en ing he occu ence o se e e hoo
lesions. Howe e , none o hese esea che s used his echnique o di ide cows wi h di e en deg ees o
locomo ion sco es.
2.4. Measu ing Gai and/o Ac i i y Using Accele ome e s
O he s udies use accele ome e s o measu e he ac i i y o gai ea u es o cows and hei ela ion
o lameness. In a s udy by Maz ie e al. [48] he educ ion in ac i i y (i.e., a e age s eps / hou s) o
lame cows anged om 9 o 68% and almos hal o he lame cows showed a educ ion o mo e han 5%
du ing he 7 o 10 days p io o clinical signs. In 92% o he lameness cases, he dec ease in ac i i y was
abo e 15%. As such de ices a e cu en ly used o oes us de ec ion in ca le, S eensels e al. [49] also
sugges ed using hese de ices al eady p esen on many a ms as a s a ing poin o u u e esea ch o
iden i y lameness p oblems. Pas ell e al. [50] used a cus om-made wi eless 3D accele ome e sys em o
measu e empo al gai cha ac e is ics on all 4 limbs o he cows. Di e ences in symme y a iance and
o wa d accele a ion we e obse ed be ween lame and non-lame cows.
Chapinal e al. [51] used i e 3D accele ome e s on cows, one on each limb and one as ened a ound
he o so. A single de ice a ached o one o he legs appea ed o be su icien o measu e he walking
speed o cows, which was associa ed wi h locomo ion sco es. In ano he s udy accele ome e s we e
moun ed on a hind leg on 348 cows in 401 lac a ions on ou comme cial a ms [52]. On hese cows, 959
locomo ion sco es we e assessed by ained echnicians, and he ac i i y le el di e ed al eady be ween
cows o locomo ion sco e 1 and sco e 2 (on a scale om 1 o 5), hus al eady de ec ing he ea ly signs o
lameness [52]. Using a senso a ached o one o he on legs o he cows o measu e ac i i y and lying
beha io , Alsaaod e al. [53] we e able o p edic lameness in cows wi h an accu acy o 76% based on
de ia ions om no mal beha io .
Animals 2015,5869
2.5. Measu ing Lying Time Using Accele ome e s
Almos a decade ago, Munksgaa d e al. [54] sugges ed he use o senso s ha measu e accele a ion
ei he in 1 o in 3 dimensions o au oma ically moni o ac i i y o he s anding and lying beha io o
cows. They ound excellen accu acies be ween he senso da a a ached o he legs o he cows and
obse a ions o lying and s anding (0.99), ac i i y (0.89) and o numbe o s eps (0.84). Since hen, a
as numbe o s udies ha e used accele ome e s o measu e dai y cow ac i i y and beha io . The esul s
o Munksgaa d e al. [54] we e con i med by a alida ion s udy o O’D iscoll e al. [55], whe e da a
logge s (1D o 3D accele ome e s) a ached o he legs o he cows measu ed ac i i y, lying and s anding
beha io s in cows wi h a conco dance o 96.3%, sugges ing ha such senso s a e a good al e na i e o
di ec beha io al obse a ions.
Simila senso s a e used o in es iga e he link be ween lying beha io o cows and lameness. The
esul s o Yun a e al. [56] sugges ha moni o ing lying beha io a ound eeding ime could help a me s
in inding lame animals, as lame cows s and up la e and lie down ea lie a e esh eed is deli e ed
compa ed o non-lame cows. Blackie e al. [2] and Calde on and Cook [57] we e able o link lameness o
di e en lying ime du a ions and numbe o lying bou s bo h be o e and a e cal ing when compa ing
non-lame and lame cows.
2.6. Combining Al eady A ailable Senso Da a
Van He em e al. [40] used senso s al eady p esen on a ms o de ec clinically lame animals wi h
a sensi i i y o 89%, a speci ici y o 85% and a co ec classi ica ion a e o 86% using he se en inpu
a iables bes co ela ed wi h lameness, i.e., he daily milk yield 4 days be o e diagnosis, he slope
coe icien o he daily milk yield 4 days be o e diagnosis, he nigh ime o day ime neck ac i i y a io 6
days be o e diagnosis, he milk yield week di e ence a io 4 days be o e diagnosis, he milk yield week
di e ence 4 days be o e diagnosis, he neck ac i i y le el du ing he day ime 7 days be o e diagnosis,
and he umina ing ime du ing nigh ime 6 days be o e diagnosis. Simila ly, Kamphuis e al. [58]
used da a om senso s al eady a ailable on a m such as li e weigh , ac i i y and milk cha ac e is ics.
A compa ison o hose da a o a ime pe iod o 14 days be o e a lameness e en wi h con ol da a o
non-lame cows esul ed in li e weigh , ac i i y and milking o de being selec ed o ha e he highes
po en ial o iden i ying he onse o lameness. De ec ion models ha combine hese h ee a iables
eached a speci ici y o abou 80%, bu he de ec ion pe o mance emained low (50%).
Ac i i y da a in e ms o lying beha io , combined wi h milk yield and eeding da a in he o m o
concen a e le -o e s in he milking obo , we e used o de ec lameness by de Mol e al. [59] wi h an
o e all sensi i i y o 86% and speci ici y o 89%. Libe a i and Zappa igna [60] combined measu emen s
o milk p oduc ion, milk low and animal ac i i y o de ec ion o abno mal cow heal h. Howe e ,
hei models we e no accu a e enough (sensi i i y < 56%) o de ec ing lameness p oblems in cows.
Ga cia e al. [61] combined 320 a iables ob ained om a milking obo and neck-based ac i i y me e s.
Eigh y-eigh cows we e epea edly locomo ion sco ed on a scale om 1 o 4, hen dicho omized in o
non-lame (sco e 1) and clinically lame (sco es 3 and 4), hus omi ing sco e 2 om analysis. A model
o i s pa i y cows and mul ipa ous cows we e buil , which achie ed sensi i i y and speci ici y alues
a ound 80%. A he op imum alues in he ecei e ope a ing cha ac e is ic cu e, he alse-posi i e a e
Animals 2015,5876
Table 5. Con .
Expe imen al se -up Lameness e e ence
Sou ce Senso ype Numbe
o cows
No mal
ou ine
N˝o a iables pe
a iable ype
Au oma ed
measu emen s
In
eal- ime
Pe son
pe o ming
sco ing
Numbe
classes
used
Cu -o le els
o lameness
Miekley e al. [69]
pedome e ac i i y
315 yes
ac i i y da a (1)
yes no
he dsmen
and
e e ina ian
Logbook
lameness
e en s
eeding pa e ns eeding da a (3)
milking da a (1) milking da a (1)
eeding da a (1) eeding da a (1)
Kamphuis
e al. [58]
weigh scales 318 lame
and 3180
non-lame
yes
weigh scales (1)
yes
yes ( ained)
he dsmen 1 o 5
Logbook
lameness
e en s
pedome e s pedome e s (1) no
milk me e s milk me e s (4) yes
Van He em
e al. [40]
neck ac i i y 44 lame
and 74
non-lame
yes
neck ac i i y (1)
yes
no
he dsmen
Logbook
lameness
e en s
umina ing ime umina ing ime (1) no
milking da a milking da ad (5) yes
Ga cia e al. [61]milking da a neck
ac i i y 88 yes milking da a (320)
ac i i y index no no obse e 1 o 4
non-lame: 1
lame: 3 + 4
(sco e 2
excluded)
No ing e al. [70]
au oma ic eede s
milking da a
weigh scales
50 yes
Feeding beha io
(4) milk yield and
milking equency
body weigh
yes no obse e 1 o 5
non-lame: 1 +
2 mildly lame:
3 se e ely
lame: 4 + 5
No mal ou ine: cows walk in no mal ou ine o we e guided in an expe imen al se up; Au oma ed: measu emen s we e au oma ed wi hou p esence o an ope a o ;
Real- ime: esul s o a iables a e a ailable in eal- ime.

Animals 2015,5877
Finally, i da a and lameness classi ica ion esul s we e o be a ailable in eal- ime, cows in need o
ea men could be sepa a ed om he he d au oma ically. Ob iously, such a p ocedu e would equi e
ba n equipmen whe e he cows can be iden i ied and hen guided o a sepa a e a ea (e.g., a e milking).
In h ee ou o ou s udies, he a iables we e calcula ed a e wa ds, and calcula ions we e no always
pe o med ully au oma ed. This means some addi ional in eg a ion wo k would be needed o ob ain a
eal- ime au oma ic lameness de ec ion sys em.
Need o au oma ed and con inuous measu emen s? Au oma ed measu emen s can ga he da a
con inuously, such ha cows can be moni o ed on a daily basis. In addi ion, hei majo ad an age
is he lack o need o he ding he cows. As cows ha e a s oic na u e, guiding hem can bias he
measu emen s, because hey will y o hide hei weakness and pain compa ed o measu emen s du ing
no mal ou ine wi hou he p esence o a human o p eda o [71]. Ideally, measu emen s should he e o e
be ully au oma ed, so no in e e ence o an ope a o is necessa y. Based on he esea ch in Table 1
o Table 5, only 63% o he s udies measu ed cows du ing no mal a ming ou ine wi hou addi ional
guiding. In s udies using o - he-shel accele ome e s o comme cially a ailable senso s, ou ou o i e
measu emen s we e pe o med wi hou guiding he cows, and in abou 82% o hese s udies, expe imen s
we e conduc ed au oma ically.
Need o ‘ea ly’ de ec ion? Th oughou he pas 20 yea s o esea ch on dai y cow lameness and
au oma ic lameness de ec ion sys ems, scien is s ha e claimed ha ea ly de ec ion o lameness signs
is bene icial in ea ing a ec ed animals be o e he p oblem becomes oo se e e. In doing so, long
las ing and cos ly ea men s, p oduc ion losses and educed wel a e long e m can be a oided. F om
a heo e ical poin o iew his migh indeed be ue, bu implemen ing he concep o “ea ly lameness
de ec ion” in p ac ice poses easonable ques ions: Wha is ea ly de ec ion? When is i needed o when
does a a me pe cei e i as an added alue? Nowadays, he bigges challenge o any lameness de ec ion
sys em is he de ec ion o ea ly onse . The de ini ion o ‘ea ly de ec ion’, howe e , is no so clea , as
‘ea ly’ can be conside ed in se e al ways.
A i s app oach o ea ly de ec ion is ‘de ec ion be o e he isual clinical signs o lameness a e
p esen ’. E en hough his is he mos logical in e p e a ion, his migh no be ele an o a lameness
de ec ion ool on comme cial a ms. I can be expec ed ha a me s will no s a ea men be o e
ob ious signs o lameness a e p esen . Acco ding o a s udy o Alawneh e al. [72], o e 65% o he
mildly lame cows and mo e han 40% o he se e ely lame cows we e ea ed mo e han h ee weeks
a e de ec ion. One o he easons why a me s pos pone ea men o pay less a en ion o lame cows
in gene al is he wo k load a peak imes, such us sowing ime in sp ing o ha es ing ime in la e
summe o au umn. The a me s’ p io i ies simply shi o o he asks [73]. As se e al hoo lesions
ake some ime o de elop, i migh also be di icul o hoo imme s o e e ina ians o de ine he
lesion causing he lameness and gi e accu a e ea men a such ea ly s ages. Iden i ica ion o p oblems
in hoo shape, on he o he hand, migh p o ide in o ma ion o he op imal iming o apply p e en i e
hoo imming. Howe e , ea ly de ec ion o lameness is a p e equisi e o e ec i e ea men , which
again may p e en he lameness om becoming ch onic [74,75]). The e o e, ano he way o app oach
‘ea ly’ de ec ion migh be ‘be o e he lame cow would be no iced by he a me ’ o e en ‘a he same
ime’. Hence, au oma ed lameness de ec ion would mo e o less be an au oma ion o he daily isual
Animals 2015,5878
locomo ion sco ing. Such an app oach o ea ly de ec ion seems mos easible o any o he de ec ion
sys ems o lameness and s ill i s he scope o suppo ing he a me in moni o ing he indi idual cows.
Need o cus om-made de ec ion sys ems? Ea ly de ec ion o lameness should be used only i i is
ele an o he a me , which sugges s ha he e may be a need o cus om-made de ec ion sys ems. Fo
a a me wi h low he d lameness p e alence and a good gene al lameness managemen , ea ly de ec ion
o new cases o lameness o mildly lame cows migh c ea e an added alue. Howe e , such a me s a e
a e. Mos a me s hugely unde es ima e no only he p e alence o he lame cows in hei he ds [62,63],
bu also he se e i y o he lameness cases [63,76]. The majo i y o a me s p e e a de ec ion sys em
which only c ea es ala ms o he se e e cases (speci ici y o >99%), as shown o au oma ed de ec ion o
mas i is [77]. Se e e cases can be di ided in o new cases o ch onic cases (i.e., hose ha cause ecu ing
ala ms). In he nex s ep, a e p e en ion and ea men p og ams ha e success ully esul ed in a lowe
lameness p e alence a he he d, he h eshold se ings o he de ec ion sys em could be changed o also
de ec he mildly lame cows o u he dec ease he lameness p e alence o a lowe le el. In gene al, i
is e y impo an o con ince he a me o us he lameness de ec ion sys em. Many a me s may be
eluc an o ely on he judgemen o an au oma ic sys em a he han hei own. The a me ’s suspicion
migh be e en wo se o ea ly lameness de ec ion, especially i he o she canno diagnose a ea able
p oblem. In gene al, a lameness de ec ion sys em should p o ide he ideal balance be ween de ec ing
almos e e y lame cow (high sensi i i y) and ha ing as ew alse de ec ions as possible (high speci ici y).
Fa me s migh be mo e eluc an abou a high numbe o alse ale s compa ed o a missed lame cow, as
alse ale s c ea e unnecessa y labo and ime o check he cows on he ale lis . Howe e , esea ch on
he expec a ions and use o a me s o a lameness de ec ion sys em migh p o ide mo e in o ma ion on
he eques ed sensi i i y and speci ici y o lameness de ec ion sys ems.
Need o eal- ime measu emen s? I he de ec ion could be pe o med in eal- ime, i.e.,
immedia ely a e he cow was measu ed by he sys em, i becomes possible o au oma ically sepa a e
a cow ha is iden i ied as lame by he so wa e and needs ex a a en ion om he a me . Such
decisions made on he spo mus be done in less han i e seconds, equi ing echnology wi h
high-speed calcula ions.
O he measu emen app oaches shown in Table 1 o Table 5, da a we e only a ailable in eal- ime
in ewe han one ou o ou s udies. The ac ha a iables measu ed wi h accele ome e s we e no
conside ed eal- ime because hey we e mos ly based on a e aged alues o e a pe iod o 24 hou s migh
ha e caused his low numbe o eal- ime measu emen sys ems. Howe e , analysis o accele ome e
da a can be pe o med e.g., du ing milking, he eby allowing eal- ime analysis. No s udy on ision
echniques analyzed he images di ec ly a e he expe imen pe iod and in si u. Whe he o no
measu emen s o a lameness de ec ion sys em should be in eal- ime a his s age o he de elopmen
p ocess emains unclea . Fa me s willing and lexible enough o ea lame cows e e y day may bene i
om eal- ime measu emen s. Fa me s planning ixed weekly ea men days may no need eal- ime
sys ems because cows ha we e iden i ied lame in he pas week could be selec ed and sepa a ed on one
speci ic day. One migh also sugges ha based on he lis wi h a en ion cows, a a me should decide
which cow should be examined and hence which cows ha e o be sepa a ed om he g oup a e nex
milking. Fo he op manage s wi h smalle he ds o ewe p oblems, his app oach migh indeed c ea e
less us a ion as such a me s ha e he skills and knowledge o decide which cow should o should
Animals 2015,5879
no ecei e ea men . On he o he hand, an au oma ically sepa a ed cow would o ce he a me o
a end o he sepa a ed cow be o e eleasing he back in o he ba n. This would enhance he chance o
ha cow ge ing p ope ea men . Sys ems wo king in eal- ime always ha e he ad an age o decisions
being made in eal- ime wi hou being a necessi y. Real- ime da a can also be s o ed and p ocessed la e
depending on he applica ion and he a me ’s eques . All op ions emain possible. Whe he o no
lameness de ec ion will be a eal- ime solu ion may also depend on in e es ed companies and wha hey
en ision as a comme cial applica ion ha can be sold o hei clien s a a easonable p ice.
A ailable space. In p ac ice, no ee space is a ailable in dai y ba ns o ins all any lameness de ec ion
sys em. This migh c ea e d awbacks o hose senso echnologies ha need an alley se -up whe e
measu emen s a e pe o med (walk-o e de ices, e.g., he Gai wise sys em, S epMe ixTM) o whe e
ideo can be eco ded ( ision echniques). The e o e, measu emen sys ems ha need less space o
ha can be included in he exis ing a m in as uc u e, like measu emen s o weigh di e ences in he
milking obo o measu emen s wi h accele ome e s, migh be mo e easible in p ac ice. I no , addi ional
space could be inco po a ed in he plans o new ba ns. Howe e , c ea ing su icien space in any exis ing
ba n is challenging, especially as cows should p e e ably pass his measu emen zone daily—i possible,
a e milking—and should be iden i ied simul aneously. This equi es a ee zone inside he ba n a e
he milking pa lo , o a y o milking obo o e en a he exi o he pas u e o measu emen s du ing he
g azing season. This d awback howe e , is especially p esen in he smalle dai y a ms. In la ge dai ies,
which a e he kind o a ms which would bene i hugely om his echnology, ins alling a lameness
de ec ion sys em migh be easie . Howe e , he mos impo an hing is o ensu e good cow a ic,
especially i a sys em is ins alled a e he milking pa lo o milking obo . Cows blocking he a ea
a ound he measu emen zone will dis up he measu emen s and c ea e measu emen ailu es and migh
e en a ec he milking ou ine [32].
4. Conclusions
In spi e o he amoun o esea ch a ailable on measu emen o gai and beha io al cha ac e is ics ha
a e ele an o lameness de ec ion, no e icien au oma ed lameness de ec ion sys em is a ailable on he
ma ke ye . This e iew ocused on he disc epancy be ween he expe imen al se -ups used in he s udies,
he s age o au oma ion o he measu emen s, and he p ac ical conside a ions when implemen ing a
lameness de ec ion sys em on a m. Mos esea ch on lameness de ec ion ocuses on he de ec ion o
se e ely lame cows, o en igno ing mildly lame cows o conside ing hem non-lame. On he o he hand,
he p ac ical easibili y o also de ec ing he mildly lame cases should be in es iga ed on a ms. This
migh esul in cus om-made lameness de ec ion sys ems ha a e adjus able depending on he deg ee
and se e i y o mildly and se e ely lame cases on ha a m and he p e e ences o he a me o speci ic
cha ac e is ics o he sys em. In addi ion, se e al senso echnologies ake up qui e some space and
will need o be e y cos -e icien in o de o a me s o decide o buy and c ea e he needed space o
he ins alla ion.
Au ho Con ibu ions
This e iew was w i en by he i s au ho ; all co-au ho s made sugges ions and al e a ions o he
e iew o imp o emen .
Animals 2015,5880
Con lic s o In e es
The au ho s decla e no con lic o in e es .
Re e ences
1. Van Nu el, A.; Zwe aeghe , I.; Pluym, L.; Van Weyenbe g, S.; Tho up, V.M.; Pas ell, M.;
Sonck, B.; Saeys, W. Lameness de ec ion in dai y cows: Pa 1. How o dis inguish be ween
non-lame and lame cows based on di e ences in locomo ion o beha io ? Animals 2015.
[C ossRe ]
2. Blackie, N.; Bleach, E.; Amo y, J.; Scai e, J. Impac o lameness on gai cha ac e is ics and lying
beha iou o ze o g azed dai y ca le in ea ly lac a ion. Appl. Anim. Beha . Sci. 2011,129, 67–73.
[C ossRe ]
3. Schlage e -Tello, A.; Bokke s, E.A.; G oo Koe kamp, P.W.; Van He em, T.; Viazzi, S.;
Romanini, C.E.; Halachmi, I.; Bah , C.; Be ckmans, D.; Lokho s , K. E ec o me ging le els
o locomo ion sco es o dai y cows on in a- and in e a e eliabili y and ag eemen . J. Dai y Sci.
2014,97, 5533–5542. [C ossRe ] [PubMed]
4. A che , S.C.; G een, M.J.; Huxley, J.N. Associa ion be ween milk yield and se iel locomo ion sco e
assessmen s in UK dai y cows. J. Dai y Sci. 2010,93, 4045–4053. [C ossRe ] [PubMed]
5. Hoge een, H.; Kamphuis, C.; S eene eld, W.; Mollenho s , H. Senso s and clinical mas i is—The
ques o he pe ec ale . Senso s 2010,10, 7991–8009. [C ossRe ] [PubMed]
6. Ru en, C.J.; Vel huis, A.G.J.; S eene eld, W.; Hoge een, H. Senso s o suppo heal h managemen
on dai y a ms. J. Dai y Sci. 2013,96, 1928–1952. [C ossRe ] [PubMed]
7. Rajkondawa , P.G.; Tasch, U.; Le cou , A.M.; E ez, B.; Dye , R.M.; Va ne , M.A. A sys em o
iden i ying lameness in dai y ca le. Appl. Eng. Ag . 2002,18, 87–96. [C ossRe ]
8. Rajkondawa , U.; Le cou , A.M.; Nee chal, N.K.; Dye , R.M.; Va ne , M.A.; E ez, B.; Tasch, U.
The de elopmen o an objec i e lameness sco ing sys em o dai y he ds: pilo s udy. T ans.
ASABE 2002,45, 1123–1125. [C ossRe ]
9. Rajkondawa , P.G.; Liu, M.; Dye , R.M.; Nee chal, N.K.; Tasch, U.; Le cou , A.M.; E ez, B.;
Va ne , M.A. Compa ison o models o iden i y lame cows based on gai and lesion sco es, and
limb mo emen a iables. J. Dai y Sci. 2006,89, 4267–4275. [C ossRe ]
10. Bicalho, R.C.; Cheong, S.H.; C ame , G.; Gua d, C.L. Associa ion be ween a isual and an
au oma ed locomo ion sco e in lac a ing Hols ein cows. J. Dai y Sci. 2007,90, 3294–3300.
[C ossRe ] [PubMed]
11. Liu, J.; Nee chal, N.K.; Tasch, U.; Dye , R.M.; Rajkondawa , P.G. Enhancing he p edic ion
accu acy o bo ine lameness models h ough ans o ma ions o limb mo emen a iables. J. Dai y
Sci. 2009,92, 2539–2550. [C ossRe ] [PubMed]
12. Liu, J.; Dye , R.M.; Nee chal, N.K.; Tasch, U.; Rajkondawa , P.G. Di e si y in he magni ude o
hind limb unloading occu s wi h simila o ms o lameness in dai y cows. J. Dai y Res. 2011,78,
1–10. [C ossRe ] [PubMed]
13. Schamha d , H.C.; Vandenboge , A.J.; Ha man, W. Measu emen Techniques in Animal
Locomo ion Analysis. Ac a Ana omica 1993,146, 123–129. [C ossRe ] [PubMed]
Animals 2015,5881
14. Sco , G.B. Changes in limb loading wi h lameness o a numbe o iesian ca le. B i . Ve . J.
1989,145, 28–38. [C ossRe ]
15. Skjø h, F.; Tho up, V.M.; do Nascimen o, O.F.; Ing a sen, K.L.; Rasmussen, M.D.; Voig , M.
Compu e ized iden i ica ion and classi ica ion o s ance phases as made by on o hind ee o
walking cows based on 3-dimensional g ound eac ion o ces. Compu . Elec . Ag . 2013,90,
7–13. [C ossRe ]
16. Tho up, V.M.; do Nascimen o, O.F.; Skjø h, F.; Voigh , M.; Rasmussen, M.D.; Bennedsgaa d, T.W.;
Ing a sen, K.L. Changes in gai symme y in heal hy and lame dai y cows based on 3-dimensional
g ound eac ion o ces ollowing claw imming. J. Dai y Sci. 2014,97, 7679–7684. [C ossRe ]
[PubMed]
17. Pas ell, M.; Hau ala, M.; Poikalainen, V.; P aks, J.; Vee mae, I.; Kujala, M.; Ahokas, J. Au oma ic
obse a ion o cow leg heal h using load senso s. Compu . Elec . Ag ic. 2008,62, 48–53.
[C ossRe ]
18. Pas ell, M.; Takko, H.; G ohn, H.; Hau ala, M.; Poikalainen, V.; P aks, J.; Vee mae, I.; Kujala, M.;
Ahokas, J. Assessing cows’ wel a e: Weighing he cow in a milking obo . Biosys . Eng. 2006,93,
81–87. [C ossRe ]
19. Pas ell, M.E.; Kujala, M. A p obabilis ic neu al ne wo k model o lameness de ec ion. J. Dai y
Sci. 2007,90, 2283–2292. [C ossRe ] [PubMed]
20. Pas ell, M.; Madsen, H. Applica ion o CUSUM cha s o de ec lameness in a milking obo .
Expe Sys . Appl. 2008,35, 2032–2040. [C ossRe ]
21. Kujala, M.; Pas ell, M.; So e i, T. Use o o ce senso s o de ec and analyse lameness in dai y
cows. Ve . Rec. 2008,162, 365–368. [C ossRe ] [PubMed]
22. Ne eux, S.; Wea y, D.M.; Rushen, J.; on Keyse lingk, M.A.G.; de Passille, A.M. Hoo discom o
changes how dai y ca le dis ibu e hei body weigh . J. Dai y Sci. 2006,89, 2503–2509.
[C ossRe ]
23. Chapinal, N.; de Passille, A.M.; Rushen, J.; Wagne , S. E ec o analgesia du ing hoo imming on
gai , weigh dis ibu ion, and ac i i y o dai y ca le. J. Dai y Sci. 2010,93, 3039–3046. [C ossRe ]
[PubMed]
24. Chapinal, N.; de Passille, A.M.; Rushen, J.; Wagne , S. Au oma ed me hods o de ec ing lameness
and measu ing analgesia in dai y ca le. J. Dai y Sci. 2010,93, 2007–2013. [C ossRe ] [PubMed]
25. Pas ell, M.; Hanninen, L.; de Pasille, A.M.; Rushen, J. Measu es o weigh dis ibu ion o dai y
cows o de ec lameness and he p esence o hoo lesions. J. Dai y Sci. 2010,93, 954–960.
[C ossRe ] [PubMed]
26. Chapinal, N.; Tucke , C.B. Valida ion o an au oma ed me hod o coun s eps while cows s and on
a weighing pla o m and i s applica ion as a measu e o de ec lameness. J. Dai y Sci. 2012,95,
6523–6528. [C ossRe ] [PubMed]
27. an de Tol, P.P.J.; Me z, J.H.M.; Noo dhuizen-S assen, E.N.; Back, W.; B aam, C.R.; Weijs, W.A.
The p essu e dis ibu ion unde he bo ine claw du ing squa e s anding on a la subs a e. J. Dai y
Sci. 2002,85, 1476–1481. [C ossRe ]

Animals 2015,5882
28. an de Tol, P.P.J.; Me z, J.H.M.; Noo dhuizen-S assen, E.N.; Back, W.; B aam, C.R.; Weijs, W.A.
The e ical g ound eac ion o ce and he p essu e dis ibu ion on he claws o dai y cows while
walking on a la subs a e. J. Dai y Sci. 2003,86, 2875–2883. [C ossRe ]
29. Mae ens, W.; Vangey e, J.; Bae , J.; Jan uan, A.; Me ens, K.C.; De Campenee e, S.; Pluk, A.;
Opsome , G.; Van Weyenbe g, S.; Van Nu el, A. De elopmen o a eal ime cow gai acking
and analysing ool o assess lameness using a p essu e sensi i e walkway: The GAITWISE sys em.
Biosys . Eng. 2011,110, 29–39. [C ossRe ]
30. Van Nu el, A.; Sp enge , M.; Tuy ens, F.; Mae ens, W. Can people see gai i egula i ies? Anim.
Wel . 2009,19, 433–439.
31. Van Nu el, A.; Vangey e, J.; Me ens, K.C.; Pluym, L.; De Campenee e, S.; Saeys, W.;
Opsome , G.; Van Weyenbe g, S. Explo a ion o measu emen a ia ion o gai a iables o ea ly
lameness de ec ion in ca le using he GAITWISE. Li es ock Sci. 2013,156, 88–95. [C ossRe ]
32. Van Nu el, A. De elopmen o an au oma ed de ec ion sys em o lameness in ca le: The Gai wise
sys em. Ph.D. Thesis, Ka holic Uni e si y o Leu en, Leu en, Belgium, 2014; p. 209.
33. Flowe , F.C.; Sande son, D.J.; Wea y, D.M. Hoo pa hologies in luence kinema ic measu es o dai y
cow gai . J. Dai y Sci. 2005,88, 3166–3173. [C ossRe ]
34. Song, X.Y.; Le oy, T.; V anken, E.; Mae ens, W.; Sonck, B.; Be ckmans, D. Au oma ic de ec ion
o lameness in dai y ca le—Vision-based ackway analysis in cow's locomo ion. Compu . Elec .
Ag ic. 2008,64, 39–44. [C ossRe ]
35. Pluk, A.; Bah , C.; Le oy, T.; Pou sabe i, A.; Song, X.; V anken, E.; Mae ens, W.; Van Nu el, A.;
Be ckmans, D. E alua ion o s ep o e lap as an au oma ic measu e in dai y cow locomo ion. T ans.
ASABE 2010,53, 1305–1312. [C ossRe ]
36. Pluk, A.; Bah , C.; Pou sabe i, A.; Mae ens, W.; Van Nu el, A.; Be ckmans, D. Au oma ic
measu emen o ouch and elease angles o he e lock join o lameness de ec ion in dai y ca le
using ision echniques. J. Dai y Sci. 2012,95, 1738–1748. [C ossRe ] [PubMed]
37. Pou sabe i, A.; Bah , C.; Pluk, A.; Van Nu el, A.; Be ckmans, D. Real- ime au oma ic lameness
de ec ion based on back pos u e ex ac ion in dai y ca le: Shape analysis o cow wi h image
p ocessing echniques. Compu . Elec . Ag ic. 2010,74, 110–119. [C ossRe ]
38. Pou sabe i, A.; Bah , C.; Pluk, A.; Be ckmans, D.; Vee mäe, I.; Kokin, E.; Pokalainen, V. Online
lameness de ec ion in dai y ca le using Body Mo emen Pa e n (BMP). In P oceedings o he 11 h
In e na ional Con e ence on In elligen Sys ems Designs and Applica ions, Co doba, Spain, 22–24
No embe 2011; pp. 732–736.
39. Viazzi, S.; Bah , C.; Schlage e -Tello, A.; Van He em, T.; Romanini, C.E.B.; Pluk, A.;
Halachmi, I.; Lokho s , C.; Be ckmans, D. Analysis o indi idual classi ica ion o lameness using
au oma ic measu emen o back pos u e in dai y ca le. J. Dai y Sci. 2013,96, 257–266. [C ossRe ]
[PubMed]
40. Van He em, T.; Mal z, E.; An le , A.; Romanini, C.E.B.; Viazzi, S.; Bah , C.; Schlage e -Tello, A.;
Lokho s , C.; Be ckmans, D.; Halachmi, I. Lameness de ec ion based on mul i a ia e con inuous
sensing o milk yield, umina ion, and neck ac i i y. J. Dai y Sci. 2013,96, 4286–4298. [C ossRe ]
[PubMed]
Animals 2015,5883
41. Pou sabe i, A.; Bah , C.; Pluk, A.; Mae ens, W.; Vee mäe, I.; Kokin, E.; P aks, J.; Pokalainen, V.;
Pas ell, M.; Ahokas, J.; Van Nu el, A.; Vangey e, J.; Sonck, B.; Be ckmans, D. Image based
sepa a ion o dai y cows o au oma ic lameness de ec ion wi h a eal ime ision sys em. In
P oceedings o ASABE Annual In e na ional Mee ing, S Joseph, NE, USA, 21–24 June 2009;
p. 10.
42. Viazzi, S.; Bah , C.; Van He em, T.; Schlage e -Tello, A.; Romanini, C.E.B.; Halachmi, I.;
Lokho s , C.; Be ckmans, D. Compa ison o a h ee-dimensional and wo-dimensional came a
sys em o au oma ed measu emen s o back pos u e in dai y cows. Compu . Elec . Ag ic. 2014,
100, 139–147. [C ossRe ]
43. Van He em, T.; Viazzi, S.; S eensels, M.; Mal z, E.; An le , A.; Alchana is, V.;
Schlage e -Tello, A.; Lokho s , C.; Romanini, C.E.B.; Bah , C.; Be ckmans, D.; Halachmi, I.
Au oma ic lameness de ec ion based on consecu i e 3D- ideo eco dings. Biosys . Eng. 2014,
119, 108–116. [C ossRe ]
44. Eddy, A.L.; Van Hoogmoed, L.M.; Snyde , J.R. The ole o he mog aphy in he managemen o
equine lameness. Ve . J. 2001,162, 172–181. [C ossRe ] [PubMed]
45. Nikkhah, A.; Plaizie , J.C.; Eina son, M.S.; Be y, R.J.; Sco , S.L.; Kennedy, A.D. Sho
communica ion: In a ed h emog aphy and isual examina ion o hoo es o dai y cows in wo
s ages o lac a ion. J. Dai y Sci. 2005,88, 1749–1753. [C ossRe ]
46. Alsaaod, M.; Busche , W. De ec ion o hoo lesions using digi al in a ed he mog aphy in dai y
cows. J. Dai y Sci. 2012,95, 735–742. [C ossRe ] [PubMed]
47. S okes, J.E.; Leach, K.A.; Main, D.C.J.; Whay, H.R. An in es iga ion in o he use o in a ed
he mog aphy (IRT) as a apid diagnos ic ool o oo lesions in dai y ca le. Ve . J. 2012,193,
674–678. [C ossRe ] [PubMed]
48. Maz ie , H.; Tal, S.; Aizinbud, E.; Ba gai, U. A ield in es iga ion o he use o he pedome e o
he ea ly de ec ion o lameness in ca le. Can. Ve . J. 2006,47, 883–886. [PubMed]
49. S eensels, M.; Bah , C.; Be ckmans, D.; Halachmi, I.; An le , A.; Mal z, E. Lying pa e ns o high
p oducing heal hy dai y cows a e cal ing in comme cial he ds as a ec ed by age, en i onmen al
condi ions and p oduc ion. Appl. Anim. Beha . Sci. 2012,136, 88–95. [C ossRe ]
50. Pas ell, M.; Tiusanen, J.; Hakoja i, M.; Hanninen, L. A wi eless accele ome e sys em wi h
wa ele analysis o assessing lameness in ca le. Biosys . Eng. 2009,104, 545–551. [C ossRe ]
51. Chapinal, N.; de Passille, A.M.; Pas ell, M.; Hanninen, L.; Munksgaa d, L.; Rushen, J.
Measu emen o accele a ion while walking as an au oma ed me hod o gai assessmen in dai y
ca le. J. Dai y Sci. 2011,94, 2895–2901. [C ossRe ] [PubMed]
52. Tho up, V.M.; Munksgaa d, L.; Robe , P.-E.; E ha d, H.W.; Thomsen, P.T.; F iggens, N.C.
Lameness de ec ion ia leg-moun ed accele ome e s on dai y cows on ou comme cial a ms.
Animal 2015. [C ossRe ] [PubMed]
53. Alsaaod, M.; Rome , C.; Kleinmanns, J.; Hend iksen, K.; Rose-Meie ho e , S.; Plume , L.;
Busche , W. Elec onic de ec ion o lameness in dai y cows h ough measu ing pedome ic ac i i y
and lying beha io . Appl. Anim. Beha . Sci. 2012,142, 134–141. [C ossRe ]
54. Munksgaa d, L.; Reenen, C.G.; Boyce, R. Au oma ic moni o ing o lying, s anding and walking
beha io in dai y ca le. J. Anim. Sci. 2006,84 (Suppl.), 304.
Animals 2015,5884
55. O'D iscoll, K.; Boyle, L.; Hanlon, A. A b ie no e on he alida ion o a sys em o eco ding lying
beha iou in dai y cows. Appl. Anim. Beha . Sci. 2008,111, 195–200. [C ossRe ]
56. Yun a, C.; Guasch, I.; Bach, A. Lying beha io o lac a ing dai y cows is in luenced by lameness
especially a ound eeding ime. J. Dai y Sci. 2012,95, 6546–6549. [C ossRe ] [PubMed]
57. Calde on, D.F.; Cook, N.B. The e ec o lameness on he es ing beha io and me abolic s a us o
dai y ca le du ing he ansi ion pe iod in a ees all-housed dai y he d. J. Dai y Sci. 2011,94,
2883–2894. [C ossRe ] [PubMed]
58. Kamphuis, C.; F ank, E.; Bu ke, J.K.; Ve ke k, G.A.; Jago, J.G. Applying addi i e logis ic
eg ession o da a de i ed om senso s moni o ing beha io al and physiological cha ac e is ics
o dai y cows o de ec lameness. J. Dai y Sci. 2013,96, 7043–7053. [C ossRe ] [PubMed]
59. de Mol, R.M.; And e, G.; Bleume , E.J.B.; an de We , J.T.N.; de Haas, Y.; an Reenen, C.G.
Applicabili y o day- o-day a ia ion in beha io o he au oma ed de ec ion o lameness in dai y
cows. J. Dai y Sci. 2013,96, 3703–3712. [C ossRe ] [PubMed]
60. Libe a i, P.; Zappa igna, P. Imp o ing he au oma ed moni o ing o dai y cows by in eg a ing
a ious da a acquisi ion sys ems. Compu . Elec . Ag ic. 2009,68, 62–67. [C ossRe ]
61. Ga cia, E.; Klaas, I.; Amigo, J.M.; B o, R.; Ene oldsen, C. Lameness de ec ion challenges in
au oma ed milking sys ems add essed wi h pa ial leas squa es disc iminan analysis. J. Dai y Sci.
2014,97, 7476–7486. [C ossRe ] [PubMed]
62. Espejo, L.A.; End es, M.I.; Sal e , J.A. P e alence o lameness in high-p oducing Hols ein cows
housed in ees all ba ns in Minneso a. J. Dai y Sci. 2006,89, 3052–3058. [C ossRe ]
63. Leach, K.A.; Whay, H.R.; Maggs, C.M.; Ba ke , Z.E.; Paul, E.S.; Bell, A.K.; Main, D.C.J. Wo king
owa ds a educ ion in ca le lameness: 1. Unde s anding ba ie s o lameness con ol on dai y
a ms. Res. Ve . Sci. 2010,89, 311–317. [C ossRe ] [PubMed]
64. Blackie, N.; Bleach, E.; Amo y, J.; Scai e, J. Associa ion be ween locomo ion sco e and kinema ic
measu es in dai y cows wi h a a ying hoo lesion ype. J. Dai y Sci. 2013,96, 3564–3572.
[C ossRe ] [PubMed]
65. I o, K.; on Keyse lingk, M.A.G.; LeBlanc, S.J.; Wea y, D.M. Lying beha io as an indica o o
lameness in dai y cows. J. Dai y Sci. 2010,93, 3553–3560. [C ossRe ] [PubMed]
66. Na a o, G.; G een, L.E.; Tadich, N. E ec o lameness and lesion speci ic causes o lameness on
ime budge s o dai y cows a pas u e and when housed. Ve . J. 2013,197, 788–793. [C ossRe ]
[PubMed]
67. K ame , E.; Ca e o, D.; S ame , E.; K ie e , J. Mas i is and lameness de ec ion in dai y cows by
applica ion o uzzy logic. Li es ock Sci. 2009,125, 92–96. [C ossRe ]
68. Miekley, B.; T aulsen, I.; K ie e , J. De ec ion o mas i is and lameness in dai y cows using wa ele
analysis. Li es ock Sci. 2012,148, 227–236. [C ossRe ]
69. Miekley, B.; S ame , E.; T aulsen, I.; K ie e , J. Implemen a ion o mul i a ia e cumula i e sum
con ol cha s in mas i is and lameness moni o ing. J. Dai y Sci. 2013,96, 5723–5733. [C ossRe ]
[PubMed]
70. No ing, M.; Häggman, J.; Simojoki, H.; Tamminen, P.; Winckle , C.; Pas ell, M. Lameness impai s
eeding beha iou o dai y cows. J. Dai y Sci. 2014,97, 4317–4321. [C ossRe ] [PubMed]
Animals 2015,5885
71. Wea y, D.; Huzzley, J.M.; on Keyse linck, M.A.G. Boa d in i ed e iew: Using beha io o p edic
and iden i y ill heal h in animals. J. Anim. Sci. 2009,87, 770–777. [C ossRe ] [PubMed]
72. Alawneh, J.I.; La en, R.A.; S e enson, M.A. The e ec o lameness on he e ili y o dai y ca le
in a seasonally b eeding pas u e-based sys em. J. Dai y Sci. 2011,94, 5487–5493. [C ossRe ]
[PubMed]
73. Van He em, T.; S eensels, M.; Viazzi, S.; Bah , C.; Romanini, C.E.B.; Lokho s , C.;
Schlage e -Tello, A.; Mal z, E.; Halachmi, I.; Be ckmans, D. E ec i cow a ic on an
implemen ed au oma ic 3D ision moni o o dai y cow locomo ion. In P oceedings o he 65 h
Annual Mee ing o he Eu opean Fede a ion o Animal Science, Copenhagen, Denma k, 25–29
Augus 2014.
74. Cla kson, M.J.; Downham, D.Y.; Faull, W.B.; Hughes, J.W.; Manson, F.J.; Me i , J.B.;
Mu ay, R.D.; Russell, W.B.; Su he s , J.E.; Wa d, W.R. Incidence and p e alence o lameness
in dai y ca le. Ve . Rec. 1996,138, 563–567. [C ossRe ] [PubMed]
75. Zimme man, A. Lameness in Dai y Ca le: A e Ac i i y Le els, Hoo Lesions and Lameness
Co ela ed? Facul y o Land and Food Sys ems. Ph.D. Thesis, The Uni e si y o B i ish Colombia,
Vancou e , Canada, 2001; p. 32.
76. B uijnis, M.; Hoge een, H.; Ga o h, C.; S assen, E. Dai y a me s’ a i udes and in en ions
owa ds imp o ing dai y cow oo heal h. Li es ock Sci. 2013,155, 103–113. [C ossRe ]
77. Mollenho s , H.; Rijkaa , L.J.; Hoge een, H. Mas i is ale p e e ences o a me s milking wi h
au oma ic milking sys ems. J. Dai y Sci. 2012,95, 2523–2530. [C ossRe ] [PubMed]
78. Bah , C.; Van Nu el, A.; Van Weyenbe g, S.; Be ckmans, D. The ease o mo emen —How
au oma ic gai and pos u e analysis can con ibu e o ea ly lameness de ec ion in dai y ca le.
In Li es ock Housing 22. Mode n Managemen o Ensu e Op imal Heal h and Wel a e o Fa m
Animals; Aland, A., Banhazi, T., Eds.; Wageningen Academic Publishe s: Wageningen, The
Ne he lands, 2014; pp. 419–430.
© 2015 by he au ho s; licensee MDPI, Basel, Swi ze land. This a icle is an open access a icle
dis ibu ed unde he e ms and condi ions o he C ea i e Commons A ibu ion license
(h p://c ea i ecommons.o g/licenses/by/4.0/).