JOURNAL OF SPORTS SCIENCES

2018, VOL. 36, NO. 19, 2155–2163

https://doi.org/10.1080/02640414.2018.1432066

Spatiotemporal analysis of 3D kinematic asymmetry in professional cycling during

an incremental test to exhaustion
Camille Pouliquen a, Guillaume Nicolasa, Benoit Bideaua, Gérard Garob, Armand Megretc, Paul Delamarchea
and Nicolas Bideaua
a
M2S Laboratory, UFR STAPS, University Rennes 2, ENS Rennes, Bruz, France; bBrest Sport Medical Center, Brest, France; cFrench Cycling Federation,
Montigny le Bretonneux, France

ABSTRACT ARTICLE HISTORY

This study investigated the influence of an incremental exercise on bilateral asymmetry through the Accepted 18 December 2017
spatio-temporal evolution of 3D joint angular displacement, using the Normalized Symmetry Index (NSI)
KEYWORDS
and cross-correlation methods. Twelve professional cyclists performed an incremental test to exhaus- Kinematic asymmetry;
tion, during which motion capture was used. Results revealed a decrease in range of motion between cycling; incremental test
the first and last stages for twelve of the eighteen joint rotations, with the highest impact observed for
right hip flexion/extension (61.8 ± 4.7° to 58.8 ± 4.1°, p < 0.05, ES = 0.68). For both stages, significant
bilateral differences greater than 10° were observed for hip and knee flexion/extension (p < 0.05,
ES>0.90) and ankle and hip internal/external rotation (p < 0.05, ES>0.25). Cross-correlation displayed the
lowest pattern similarities for hip abduction/adduction and the highest similarities for knee flexion/
extension, ankle plantarflexion/dorsiflexion and hip internal/external rotation. The cross-correlation
method showed that the right leg was mostly ahead of time with respect to the left leg, a trend that
was accentuated with power output increase. Instantaneous NSI fluctuated up to 18% throughout the
pedalling cycle, with different behaviour between the power and recovery phases. This study demon-
strated the workload effects on side-to-side joint angular pattern similarity.

Introduction exercise. However, a planar assumption may be questionable

in the light of the results of Umberger and Martin (2001) that
Pedalling has usually been assessed while assuming symmetry
reported errors between 2D and 3D calculations that could
in force production and kinematics. However, cyclists can
reach 32° for maximum hip angle in the sagittal plane. The
exhibit asymmetry (Carpes, Mota, & Faria, 2010), which is
planar assumption also proves to be restrictive in that the
often associated with lower efficiency, an increased risk of
degrees of freedom (DOF) out of the sagittal plane have
overuse injuries or premature fatigue (e.g., Bini, Diefenthaler,
been suggested as a cause of overuse knee injuries (Bailey,
Carpes, & Mota, 2007; Smak, Neptune, & Hull., 1999). Several
Maillardet, & Messenger, 2003; Gregersen & Hull, 2003).
studies have quantified pedalling asymmetry using kinetic and
However, few studies have assessed the effects of workloads
kinematic analyses. Evaluating bilateral kinetic asymmetry by
on 3D kinematics. Recently, Bini et al. (2016) reported no
using devices that compare right-to-left torque and power
workload effects on 3D joint angles, although (Fang,
output (PO) has recently gained popularity (e.g., Bini &
Fitzhugh, Crouter, Gardner, & Zhang, 2016) showed that both
Hume, 2014). Some studies have focused more specifically
lateral and transverse trunk-pelvic angle range of motion
on the 2D forces generated on the pedals (Bini & Hume,
(ROM) as well as knee abduction ROM increased with work-
2014), crank torque (Carpes, Rossato, Faria, & Mota, 2007a)
load. Since cyclists adapt their joint kinematics to overcome
and external work (Daly & Cavanagh, 1976) generated during
higher workloads and progressive fatigue (Bini et al., 2010; Bini
a pedalling exercise. Discrepancies about the effect of PO on
& Diefenthaler, 2010; Dingwell, Joubert, Diefenthaeler, &
asymmetry have been reported: increased asymmetry (Bini &
Trinity, 2008; Edeline, Polin, Tourny-Chollet, & Weber, 2004;
Hume, 2014), decreased asymmetry (Carpes, Rossato, Faria, &
Sayers, Tweddle, Every, & Wiegand, 2012), some authors have
Mota, 2007b) and no asymmetry (Bini et al., 2007; Garcia-
focused on the effect of an incremental test on kinematic
Lopez, Diez-Leal, Larrazabal, & Ogueta-Alday, 2015) have
bilateral symmetry. Thus, Edeline et al. (2004) found symmetry
been reported with varying PO. As an attempt to better under-
in pedalling kinematics to be reduced as the workload
stand cycling technique, different authors have suggested _ 2max .
increased from 55% to 75% of VO
analysing joint kinematics and kinetics during an incremental
To quantify asymmetry, algebraic indexes are classically
exercise to exhaustion using a planar assumption (Bini,
used (Carpes et al., 2007a; Smak et al., 1999). The symmetry
Diefenthaeler, & Mota, 2010; Black, Sanderson, & Henning,
index (SI) and symmetry ratio (SR) evaluate bilateral differ-
1993). Bini et al. (2010) showed that cyclists modify their
ences of discrete variables (e.g., at maximal torque), such as
technique in the sagittal plane during exhaustive incremental

CONTACT Camille Pouliquen pouliquencamille@gmail.com M2S laboratory Avenue Robert Schuman, Bruz 35170, France
© 2018 Informa UK Limited, trading as Taylor & Francis Group
2156 C. POULIQUEN ET AL.

joint angular displacements/velocities, pedal forces or crank measurements. After a 4 min warm up at 100 W, the PO was
torques, but fail to analyse the motion over the entire motion increased every 2 min at 50 W increments until the cyclists’
cycle (Crenshaw & Richards, 2006). Moreover, algebraic voluntary exhaustion. PO was measured using a SRM system
indexes may lead to inconsistent results such as artificially (Schoberer Rad Meßtechnik, Science version, Germany) cali-
inflated values (Zifchock, Davis, Higginson, & Royer, 2008) brated according to the manufacturer’s recommendations.
that could exceed 13,000% during walking (Herzog, Nigg, The SRM power control provided PO feedback for each parti-
Read, & Olsson, 1989). Hence, care should be taken when cipant during the test.
interpreting this ratio calculated for discrete time events, as
it does not always provide a relevant indicator of the overall
pattern. Furthermore, these methods do not account for time Motion capture protocol
shifts between the legs, which can be addressed through
Cyclists were equipped with 42 markers placed on anatomical
waveform analysis. Thus, cross-correlation (CC) is a useful
landmarks (Figure 1) to calculate the lower limb anatomical
method for studying the patterns of rhythmic motion and
rotations following the ISB recommendations (Wu &
has been previously used in cycling to quantify the relative
Cavanagh, 1995; Wu et al., 2002). Top dead centre (TDC) and
changes in the temporal characteristics of EMG activity (Hug &
bottom dead centre (BDC) were determined during post-pro-
Dorel, 2009; Li & Caldwell, 1999). From a kinematic viewpoint,
cessing using one marker on each pedal spindle. Twelve Vicon
this method has been employed to analyse body segment
cameras (Oxford Metrics, Oxford, UK) operating at 100 Hz were
coupling by means of characteristic delays across joints during
used to capture the marker trajectories. The 3D coordinates
gait (Park, Dankowicz, & Hsiao-Wecksler, 2012) or alterations in
were smoothed using a 2nd-order Butterworth low-pass filter
foot strike patterns during running (Pohl, Messenger, &
with a cut-off frequency of 10 Hz (Yu, Gabriel, Noble, & An,
Buckley, 2006).
1999). Kinematic data were averaged over 20 crank cycles
Recently, Gouwanda and Senanayake (2011) introduced the
1 minute after the beginning of each stage.
Normalized Symmetry Index (NSI) to provide information on
the timing and magnitude of the symmetry difference
between limbs based on time histories of angular rate. The Joint angular displacements and ROM
main advantage of this approach lies in the amplitude normal- ROM was computed for each leg. The present analysis
ization that is applied to eliminate artificial inflation, which involved 9 anatomical DOFs: hip, knee flexion/extension (fle./
enables intra-cycle asymmetry analysis. ext.); hip, knee abduction/adduction (abd./add.); hip, knee and
Previous works have showed the influence of increased ankle internal/external rotation (IR/ER), ankle plantarflexion/
workload on asymmetry, but to the best of our knowledge, dorsiflexion (pla./dor.) and ankle inversion/eversion (inv./eve.).
no study has reported the influence of intensity on asymmetry Beyond bilateral ROM comparisons, the cross-correlation
with a continuous parameter within the crank cycle during an technique (Li & Caldwell, 1999) and the NSI adapted from
incremental test. Surprisingly, few studies have compared 3D Gouwanda and Senanayake (2011) were used to analyse 3D
side-to-side joint rotation patterns throughout the pedalling
cycle. Typically, it is unknown where joint kinematic asymme-
tries occur during different time intervals (e.g., power phase/
recovery phase). Indeed, asymmetry could appear at different
time points rather than at arbitrarily chosen values, such as
peak values, of the joint angle. In this context, the aim of this
study was to quantify the influence of an incremental exercise
on side-to-side 3D joint rotations and to address cycling asym-
metry via the spatio-temporal evolution of joint angular dis-
placement using the NSI and cross-correlation methods.

Methods
Twelve professional UCI continental cyclists took part in the
study. The mean (± SD) age, height, body mass and VO _ 2max
were 25.6 ± 3.8 years, 1.88 ± 0.06 m, 72.5 ± 3.5 kg,
75.2 ± 44 ml.min−1.kg−1. For all participants, the right leg
was revealed to be dominant following standard tests (De
Ruiter, De Korte, Schreven, & De Haan, 2010). All volunteers
signed an informed consent form in agreement with the local
Ethical Committee, and the study was conducted in accor-
dance with the 1975 Declaration of Helsinki.
Each cyclist performed the French Cycling Federation incre-
mental test to exhaustion with free pedalling cadence on a
SRM indoor trainer (Schoberer Rad Meßtechnik, Jülich,
Germany) customized to the participant’s own bicycle Figure 1. Maker set protocol for motion capture analysis.
 JOURNAL OF SPORTS SCIENCES 2157

joint asymmetries. For comparison with conventional alge- θmin and θmax are the minimum and the maximum angular
braic ratios, the SR (Patterson, William, Brooks, Black, & displacement, and θNSIDE is the normalized joint angle of the
McIlroy, 2010) was calculated for each time step (equation 1). considered side. Then, NSI is calculated as follows:
θL θNR  θNL
SR ¼  100 (1) NSI ¼  100% (5)
θR 0:5  ðθNR þ θNL Þ

For any given variable, the closer the NSI value is to zero, the
Cross-correlation method (CC)
more symmetric the cycle is. A positive NSI signifies a greater
Normalized cross-correlation technique was used to identify
amplitude for the right side, and vice versa. The mean
similarities between the left and right joint angle time his-
(NSImean ) and absolute values of NSI (NSIabs ) were calculated
tories (equations 2 and 3).
and allowed for comparison of bilateral intra-cycle kinematic
cθR θL ðkÞ asymmetry.
rθR θL ðkÞ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2ffi (2)
Pt¼1  2 Pt¼1 
N θR ðtÞ  θR N θL ðtÞ  θL
where
Statistical analysis

8
>
> X
Nk    X
N   
>
> θ ð Þ  θ θ ð þ Þ  θ þ θR ðtÞ  θR θL ðt  N þ kÞ  θL k 2 ½1; N
>
< R t R L t k L
t¼1 t¼Nkþ1
cθR θL ðkÞ ¼ ð3Þ
>
> X
Nk   
>
>
>
: θR ðtÞ  θR θL ðtÞ  θL k¼0
t¼1

θR and θL represent mean values for the right and left
instantaneous rotations (θR and θL ). The left joint angles
were shifted by 180° to synchronise the initial kinematic
cycle at BDC, for both legs.). τlag , expressed as pedal angle
lag, indicates the time shift of the left pattern compared to
the right pattern and corresponds to the value k for which
the correlation coefficient rθR θL ðkÞ is maximal. A positive
τlag indicates that the left pattern is shifted forward in
the crank cycle relative to the right pattern right leg is
taken as the reference. For each anatomical rotation, the
maximal cross-correlation value (rmax ) corresponding to the
best overlap between the right and left limb joint angles
was analysed.
Normalized Symmetry Index (NSI). θR and θL were normal-
ized in θNR and θNL between one and two to avoid artificial
inflation. A min-max normalization was applied following the
equation:
ROM, NSI and CC data were log transformed before analysis. A
Shapiro-Wilk normality test was employed. A Wilcoxon sign
rank test was used to compare the nonparametric results of
ROM obtained for the stages of 150 W and 350 W of each leg.
The level of significance was set at P < 0.05. The Common
Language Effect Size (Table 1) was calculated to present the
magnitude of the variables that were significantly different
after statistical analysis (Hentschke & Stüttgen, 2011). To inter-
pret the results of the effect sizes, we used the correspon-
dence table with Cohen effect size (Cohen, 1988) proposed in
the study of Dunlap (1999). The magnitude of the difference
was defined as small (ES<0.61), moderate (0.61<ES<0.69) and
large (ES>0.69).
Results
Range of motion

θSIDE  θmin For each PO, bilateral ROM differences were first analysed, and
θNSIDE ¼ þ 1 (4) the effect of PO on bilateral differences was investigated
θmax  θmin
(Figure 2).

Table 1. Effect size for each comparison of the ROM.

Left leg Right leg 150 W 350 W
(150 W vs. 350 W) (150 W vs. 350 W) (Right vs. left leg) (Right vs. left leg)
Joint DOF
Hip Flexion/Extension 0.63 0.68 0.96 0.96
Abduction/Adduction NS NS NS NS
Internal/External Rotation 0.66 0.64 0.26 NS
Knee Flexion/Extension NS NS 0.97 0.99
Abduction/Adduction NS 0.63 NS NS
Internal/External Rotation NS NS NS NS
Ankle Plantarflexion/Dorsiflexion NS NS NS NS
Inversion/Eversion NS 0.63 NS NS
Internal/External Rotation NS NS NS 0.25
2158 C. POULIQUEN ET AL.
Figure 2. Ranges of motions for hip, knee and ankle joints: bilateral comparison (R: Right, L: Left) and effect of power increase. * (P < 0.05) stand for bilateral
comparison. $ (P < 0.05) stand for the influence of PO.
Bilateral comparisons demonstrated significant differ-
ences (P < 0.05) with large effect sizes for hip (ES = 0.96)
and knee fle./ext. (ES = 0.97 for 150 W and ES = 0.99 for
350 W). Moreover, significant bilateral differences were
exhibited in the horizontal plane in terms of hip IR/ER at
150 W with a small effect size (ES = 0.25), as well as ankle
IR/ER at 350 W with a small effect size (ES = 0.25). In regard
to the PO increase, ROMs for both legs followed similar
trends since most of the values (twelve of the eighteen
joint rotations) decreased between the 150 and 350 W
stages. A significant decrease in the hip fle./ext. with mod-
erate effect sizes for the right limb (61.8 ± 4.7° to
58.8 ± 4.1°, ES = 0.63) and left limb (49.7 ± 4.8° to
47.4 ± 5.2°, ES = 0.68) was observed between the first and
last stages. Similarly, a significant decrease in the hip IR/ER
for the right side (21.3 ± 4.8° to 18.8 ± 5.0°, ES = 0.64) and
left side (25.5 ± 4.6° to 22.3 ± 6.6°, ES = 0.66) was quanti-
fied. However, for the knee joint, the only significant
decrease was observed for the right limb abd./add.
(13.1 ± 3.4° and 11.6 ± 2.9°) with moderate effect sizes
(ES = 0.63). Finally, a significant decrease was found for
ankle inv./eve. only for the right leg (13.1 ± 5.1° to
10.9 ± 4.6°) with a moderate effect size (ES = 0.63).
Symmetry ratio (SR)
Figure 3 illustrates a typical result from the SR calculation for
hip abd./add. and IR/ER. The high peaks highlight the artificial
inflation problem, which here were associated with a right
joint angle close to zero. Thus, the SR proves to be unusable
for continuous asymmetry quantification throughout the ped-
alling cycle.
Figure 3. Example of a temporal symmetry ratio for hip abd./add. (left) and hip IR/ER (right) subject to artificial inflation.
Normalized symmetry index and cross-correlation
Cross-correlation results displayed different pattern similarities
according to the parameter rmax (Table 2). For hip and knee
fle./ext., highly symmetric behaviours were found for both the
150 and 350 W stages (rmax ¼ 0:99). Likewise, a high similarity
was observed between the right and left leg motion for ankle
pla./dor. (rmax; 150 W ¼ 0:93 and rmax; 350 W ¼ 0:95) and hip IR/
ER (rmax; 150 W ¼ 0:95 and rmax; 350 W ¼ 0:94). Moderate similar-
ity of kinematic patterns was observed for knee abd./add.
(rmax; 150 W ¼ 0:78 and rmax; 350 W ¼ 0:75), knee IR/ER
(rmax; 150 W ¼ 0:84 and rmax; 350 W ¼ 0:80), ankle IR/ER
(rmax; 150 W ¼ 0:88 and rmax; 350 W ¼ 0:82) and ankle inv./eve.
(rmax; 150 W ¼ 0:84 and rmax; 350 W ¼ 0:64). Low similarity was
observed for hip abd./add. (rmax; 150 W ¼ 0:47 and
rmax; 350 W ¼ 0:45). In general, the rmax values remained rela-
tively constant between the first and last stages, except for
ankle inv./eve.
Considering the influence of the PO increase on the angle lag
between the legs, hip and knee abd./add. exhibited the highest
values (hip: τlag; 150 W ¼ 28:5 and τlag; 350 W ¼ 24:0 ; knee:
τlag; 150 W ¼ 10:7 and τlag; 350 W ¼ 15:3 ). The same magni-
tude was observed for knee IR/ER at 350 W (τlag; 350 W ¼ 12:6 ),
whereas no delay was quantified at 150 W (τ lag; 150 W ¼ 0:8 ).
The NSImean reported in Table 2 ranged between −5.2%
for ankle IR/ER at 350 W and 4.7% for knee IR/ER at 150 W.
Figures 4, 5 and 6 display the time histories of the NSI,
which enable asymmetry fluctuations to be quantified dur-
ing the crank cycle. Non-symmetrical behaviour of the NSI
occurred between the power and recovery phases. Indeed
the greatest NSI observed during the recovery phase was
13.4% for hip abd./add. at 150 W, whereas peak NSI reached
−14.1% during the propulsive phase. Occurrence of maximal
 JOURNAL OF SPORTS SCIENCES 2159

Table 2. NSI presented as mean±std (NSImean ) and cross correlation coefficient for each stage. Each column is the comparison results of each legs between it for each
stages. Each column of the NSI is the mean of each NSI graph represented in Figure 4.
NSImean (%) Cross correlation
Joint DOF Mean ± SD rmax τ lag ð Þ
150 W 350 W 150 W 350 W 150 W 350 W
Hip Flexion/Extension 2.2 ± 1.4 2.3 ± 1.3 0.99 0.99 −0.3 −0.6
Abduction/Adduction −0.3 ± 21.7 −1.7 ± 21.9 0.47 0.45 −28.5 −24.0
Internal/External Rotation 3.6 ± 5.9 2.1 ± 7.0 0.95 0.94 1.6 3.0
Knee Flexion/Extension −0.8 ± 1.2 −0.7 ± 1.4 0.99 0.99 2.6 1.7
Abduction/Adduction 0.7 ± 14.1 1.8 ± 14.9 0.78 0.75 −10.7 −15.3
Internal/External Rotation 4.7 ± 14.2 3.5 ± 14.6 0.84 0.80 −0.8 −12.6
Ankle Plantarflexion/Dorsiflexion −2.8 ± 8.2 −1.7 ± 6.5 0.93 0.95 −3.0 −3.1
Inversion/Eversion −0.8 ± 13.1 −1.9 ± 17.1 0.84 0.64 1.5 −7.7
Internal/External Rotation −2.1 ± 10.2 −5.2 ± 13.1 0.88 0.82 4.2 −3.8

Figure 5. Mean (full line) and SD (dashed line) of Normalized Symmetry Index
Figure 4. Mean (full line) and SD (dashed line) of Normalized Symmetry Index
for each DOF, for stage 150 W (black line) and 350 W (grey line). a: knee fle./ext.;
for each DOF, for stage 150 W (black line) and 350 W (grey line). a: hip fle./ext.;
b: knee abd./add.; c: knee IR/ER.
b: hip abd./add.; c: hip IR/ER.

values of NSI depended on the joint DOF. Fle./ext. system- Discussion

atically showed the smallest values of asymmetry compared
to the values observed for the frontal and sagittal planes
(e.g., considering knee joint: NSIabs <4% for fle./ext,
NSIabs <10:6% for abd./add. and NSIabs <18:7% for IR/ER).
Concerning the effects of PO increase, during the power
phase, a decrease is NSIabs was observed in the frontal and
horizontal planes for the hip and knee. Moreover, the ankle
joint demonstrated an increase in asymmetry. During the
recovery phase, NSIabs increased for the hip and ankle in the
frontal and horizontal planes, whereas NSIabs decreased for
the knee.
This study examined the changes in kinematic asymmetry
induced by an incremental test to exhaustion. Assuming that
sub maximal workloads do not considerably impact lower limb
joint kinematics (Bini, Senger, Lanferdini, & Lopes, 2012), only
two stages (150 and 350 W) were considered. The comparison
of ROM between the legs provides interesting insights in
terms of global asymmetry. Our findings highlighted signifi-
cant side-to-side differences in the three anatomical planes
with an increasing PO.
First, in the sagittal plane, significant bilateral differences at
the hip and knee joints were found for each PO. However, the
2160 C. POULIQUEN ET AL.
Figure 6. Mean (full line) and SD (dashed line) of Normalized Symmetry Index
for each DOF, for stage 150 W (black line) and 350 W (grey line). a: Ankle pla./
dor.; b: Ankle inv./eve.; c: Ankle IR/ER.
PO increase did not affect the knee fle./ext. ROM, but the hip
joint was significantly impacted, exhibiting a decrease in the
fle./ext. ROM. This latter result was previously discussed in Bini
et al. (2010), who reported a significant decrease for the hip
fle./ext. ROM and a significant increase in the ankle pla./dor.
ROM during an incremental test.
Non-sagittal plane lower limb kinematics are rarely evalu-
ated during cycling although it may provide information
regarding potential strain injuries (Gregersen & Hull, 2003).
Only a few studies have computed the 3D rotational DOFs
involved in the pedalling motion, albeit the values reported in
our study were non-negligible out of the sagittal plane. To the
best of our knowledge, few authors have investigated the
impact of PO on 3D ROM (Bini et al., 2016; Fang et al., 2016),
but these studies did not address bilateral differences.
Whereas significant bilateral differences were observed
only in the horizontal plane (hip and ankle IR/ER), the PO
increase did not affect the non-sagittal angular displacements
in the same manner. Indeed, a significant decrease in hip IR/ER
ROM was observed for both limbs. This reduction of joint
angular displacements in the non-sagittal planes may result
in a reduced risk of overuse injuries (Bini et al., 2016;
Heiderscheit, Hamill, & Van Emmerik, 2002). Concomitantly, a
significant decrease was observed for the knee abd./add. ROM
of the right leg only, which may result in asymmetrical move-
ments. However, to date, it is still unknown how the sagittal,
frontal and transverse plane kinematics change and what
temporal inter-relation exists during an exhaustive cycling
exercise. Previous studies have suggested that these simulta-
neous kinematical modifications in various anatomical planes
may be associated with a specific neuromuscular strategy to
combat fatigue accumulation and maintain task performance
(Gates & Dingwell, 2008). In that sense, Dingwell et al. (2008)
reported changes in knee abd./add. ROM subsequent to loca-
lized muscle fatigue during exhaustive cycling. However, ROM
bilateral difference may be limited in terms of comparing side-
to-side patterns, which limitation can be overcome using CC
and NSI (Gouwanda & Senanayake, 2011).
Using the cross-correlation method, a reduction of rmax
between the first and the last stage was observed for abd./
add. and IR/ER for all joints. A reduction of similarity may
reflect a kinematic strategy in response to the workload
increase but might also be associated with fatigue mechan-
isms that cause movement deterioration during the test. This
trend is somewhat in agreement with previous papers
regarding cycling (Carpes et al., 2010) and gait (Haddad,
Van Emmerik, Whittlesey, & Hamill, 2006). Indeed, greater
added loads during gait resulted in a systematic decrease in
the cross-correlation coefficients, which indicated a lowering
of similarity between the right and left kinematic patterns
(Haddad et al., 2006). In our study, this trend can also be
observed, except for the ankle pla./dor., which showed a
slight increase in the similarity of bilateral joint angle pat-
terns when PO increased. Cross-correlation results therefore
suggest that coordination adaptations appear as an inter-
limb level in response of pattern adaptation as the workload
increases.
More generally, for each stage, rmax revealed a nearly per-
fect symmetry for fle./ext., demonstrating that each leg experi-
enced essentially the same kinematic pattern in the sagittal
plane. However, lower bilateral pattern similarity appeared in
the other planes as revealed by a very large similarity for knee
abd./add. (rmax; 150 W ¼ 0:78), and a moderate similarity for hip
IR/ER (rmax; 150 W ¼ 0:47). This result is consistent with previous
works highlighting that symmetry between the sides of the
body plays an important role in legged locomotion and jump-
ing (Lathrop-Lambach et al., 2014; Pappas & Carpes, 2012).
Indeed, hip abd./add. was revealed to be one of the more
asymmetric rotations during bilateral forward jump-landings
and gait (Lathrop-Lambach et al., 2014).
Whereas rmax provides global information regarding simi-
larity, it does not account for the time delay between right
and left limb patterns, which can be assessed through τlag .
This temporal parameter changed with the PO increase. Few
modifications were observed in terms of the hip
(τlag; 150 W ¼ 0:3 and τlag; 350 W ¼ 0:6 Þ and knee
(τlag; 150 W ¼ 2:6 and τlag; 350 W ¼ 1:7 Þ fle./ext. However, mod-
ifications were larger for hip (τ lag; 150 W ¼ 28:5 and
τlag; 350 W ¼ 24:0 Þ and knee (τlag; 150 W ¼ 10:7 and
τlag; 350 W ¼ 15:3 Þ abd./add. This result may reflect a large
hip and moderate knee abd./add. of the left side compared to
the right side, which might be associated with the lower limb
being a closed kinematic chain during cycling because the
ankle joint is constrained by the pedal. Such a behaviour has

been previously depicted in rowing by Buckeridge, Hislop,
Bull, and McGregor (2012), who showed that hip ROM was
more asymmetrical than the knee ROM. Indeed, given that
rowers are constrained at the hands, seat and feet and are
performing a closed chain activity, any asymmetrical loading
of the foot-stretchers may cause kinematic compensation
further along the kinematic chain through flexions and rota-
tions at the lumbar–pelvic joint to maintain symmetrical
motion in the sagittal plane (Buckeridge, Bull, and McGregor
(2015). In cycling, hip rotations may also compensate for knee
rotations, as depicted by a lowering of the time lag between
the first and last stages for hip abd./add. and an increase in
the time lag for knee abd./add.. However, to our knowledge,
no study has previously reported this result.
It is also important to address the perceived influence of PO on
τlag . Indeed, time delay between bilateral patterns indicated a
forward shift for five joints in favour of the right leg at 150 W.
However, the degree of synchrony was modified by the incre-
mental test since seven joint rotations presented a forward pattern
shift with respect to the left pattern at 350 W. This result showed
that the majority of right leg’s patterns were mostly ahead of time
compared to the left leg, specifically at the end of the test. This
may illustrate the impact of the PO increase on the mastering of
the redundant DOFs, and deviation from a functionally preferred
coordination, as suggested by findings for swimming (Nikodelis,
Kollias, & Hatzitaki, 2005). This may also be related to the lower
changes in the ROM obtained for the left leg or to the relatively
large differences of 11–13° between each side for hip and knee
fle./ext. Since the participants were all right-leg-dominant, the
angle lag and bilateral ROM differences mainly in favour of the
right leg may be due to leg dominance. Indeed, a previous study
reported significantly larger values of joint angles (2–3°) in favour
of the dominant leg during a running single leg jumping
(Sugiyama et al., 2014), but no data regarding 3D analysis in
relation to leg dominance in cycling is available. Among all of
the parameters quantified in the present study, τlag was the most
sensitive indicator and the most modified by PO. Surprisingly,
except for a few studies on gait (e.g., Haddad et al., 2006), cycling
asymmetry has scarcely been evaluated using the CC method.
Moreover, this method fails to provide instantaneous information
of the timing and magnitude of the symmetry difference between
the limbs, which can be evaluated through NSI.
Whereas NSImean only provides global information of asym-
metry that revealed variations less than 5.2%, it is not suffi-
cient for comparing the similarity of the right/left pattern
throughout the crank cycle. Thus, NSI plots allow specific
regions of maximal asymmetry to be identified throughout
the crank cycle by providing the timing and magnitude of
the movement deviations between left and right limbs. As
illustrated in Figures 4, 5 and 6, the peak values of asymmetry
and corresponding occurrence throughout the crank cycle
differed between the power phase and recovery phase.
Moreover, much larger NSI values were observed in frontal
and transversal planes compared to those calculated in sagit-
tal plane. Previous studies have reported the highest values of
asymmetry in the frontal and transverse planes compared to
those in the sagittal plane during gait which may be due to a
smaller solution space during linear movement for sagittal
JOURNAL OF SPORTS SCIENCES 2161
variables (Nigg, Vienneau, Maurer, & Nigg, 2013). However, it
is questionable if such a rationale may be employed in cycling.
Although a limited number of studies have provided informa-
tion on frontal and transverse plane motion during cycling
(Bailey et al., 2003; Carpes, Dagnese, Mota, & Stefanyshyn,
2009; Fang et al., 2016; Sanderson & Black, 2003), to the best
of our knowledge, only one study has explored asymmetry out
of sagittal plane in cycling (Edeline et al., 2004). However, non-
sagittal motions such as hip and knee abduction/adduction
have been suggested as a cause of overuse knee injuries in
cycling (Bailey et al., 2003; Bini et al., 2016; Gregersen & Hull,
2003) or in running (Hreljac & Ferber, 2006).
Asymmetry level clearly differs between recovery phase
and power phase when increasing PO. Indeed, during power
phase, the PO increase affects asymmetry in the non-sagittal
planes for knee and hip by decreasing NSIabs , and a concomi-
tant increase of asymmetry for the ankle is highlighted. During
the recovery phase, NSI exhibited the greatest amplitude for
the hip and ankle in frontal and horizontal planes and a
concomitant decrease in asymmetry of the knee joint was
observed when workload increased. Since PO had little impact
on sagittal plane asymmetry, it could be hypothesized that
there are compensatory motions in frontal and transverse
plane to maintain a symmetrical pedalling motion in the
sagittal plane when workload increases.
Although kinematic asymmetry modification through NSI
may reflect pedalling technique, it does not provide informa-
tion regarding concomitant differences in bilateral joint. Since
kinematic and kinetic asymmetries are usually not correlated
(Fohanno, Nordez, Smith, & Colloud, 2015), an interesting
prospect would be to simultaneously compare the level of
kinematic and kinetic asymmetries throughout the crank
cycle. Indeed, even with an equal production of pedal forces,
joint and muscle torques can differ between sides eliciting an
increase in muscle and joint overload due to differences in
pedalling kinematics (Carpes et al., 2010). In the aim of analys-
ing the potential relationship between asymmetries, perfor-
mance and musculoskeletal injuries, future investigations
should be directed towards correlating spatio-temporal kine-
matic asymmetry with kinetic and EMG data.
Conclusion
This study showed the importance of considering non-sagittal
plane asymmetry during an incremental cycling test. In contrast
to the ROM analysis, CC and NSI were shown to be powerful
approaches for assessing bilateral pattern comparisons and
intra-cycle analysis. These indicators suggested inter-limb adap-
tations of kinematic patterns as the workload increases. In that
sense, asymmetries in frontal and horizontal planes were more
affected than in sagittal plane. More specifically, the hip exhib-
ited the most important asymmetry and appeared to be the
most impacted joint during the incremental test.
Acknowledgements
The authors wish to thank the “Bretagne Séché Environnement” UCI pro
cycling Team for their participation in the present study.
2162 C. POULIQUEN ET AL.
Disclosure statement
No potential conflict of interest was reported by the authors.
Funding
This work was funded by French National Institute of Sport (INSEP) under
Grant 12-R25. C. Pouliquen was supported by PhD scholarship of the
Region Bretagne. No relevant financial relationship exists.
ORCID
Camille Pouliquen http://orcid.org/0000-0001-6943-8586
References
Bailey, M. P., Maillardet, F. J., & Messenger, N. (2003). Kinematics of cycling
in relation to anterior knee pain and patellar tendinits. Journal of Sports
Sciences, 21(8), 649–657. doi:10.1080/0264041031000102015
Bini, R., Dagnese, F., Rocha, E., Silveira, M., Carpes, F., & Mota, C. (2016).
Three-dimensional kinematics of competitive and recreational cyclists
across different workloads during cycling. European Journal of Sport
Science, 16(5), 553–559. doi:10.1080/17461391.2015.1135984
Bini, R., Diefenthaeler, F., & Mota, C. (2010). Fatigue effects on the coordi-
native pattern during cycling: Kinetics and kinematics evaluation.
Journal of Electromyography and Kinesiology, 20(1), 102–107.
doi:10.1016/j.jelekin.2008.10.003
Bini, R., & Diefenthaler, F. (2010). Kinetics and kinematics analysis of
incremental cycling to exhaustion. Sports Biomechanics, 9(4), 223–235.
doi:10.1080/14763141.2010.540672
Bini, R., Diefenthaler, F., Carpes, F., & Mota, C. B. (2007). External work
bilateral symmetry during incremental cycling exercise. XXV ISBS
Symposium, Ouro Preto, Brazil.
Bini, R., & Hume, P. A. (2014). Assessment of bilateral asymmetry in cycling
using a commercial instrumented crank system and instrumented ped-
als. International Journal of Sports Physiology and Performance, 9(5),
876–881. doi:10.1123/ijspp.2013-0494
Bini, R., Senger, D., Lanferdini, F., & Lopes, A. L. (2012). Joint kinematics
assessment during cycling incremental test to exhaustion. Isokinetics
and Exercise Science, 20(2), 99–105.
Black, A., Sanderson, D., & Henning, E. (1993). Kinematic and kinetic
changes during an incremental exercise test on a bicycle ergometer.
Paris, France: XIV International Society of Biomechanics.
Buckeridge, E., Bull, A., & McGregor, A. (2015). Biomechanical determinants
of elite rowing technique and performance. Scandinavian Journal of
Medicine and Science in Sports, 25(2), 176–183. doi:10.1111/sms.12264
Buckeridge, E., Hislop, S., Bull, A., & McGregor, A. (2012). Kinematic asym-
metries of the lower limbs during ergometer rowing. Medicine and
Science in Sports and Exercise, 44(11), 2147–2153. doi:10.1249/
MSS.0b013e3182625231
Carpes, F. P., Dagnese, F., Mota, C. B., & Stefanyshyn, D. J. (2009). Cycling
with noncircular chainring system changes the three-dimensional kine-
matics of the lower limbs. Sports Biomechanics, 8(4), 275–283.
doi:10.1080/14763140903414409
Carpes, F. P., Mota, C. B., & Faria, I. E. (2010). On the bilateral asymmetry
during running and cycling – A review considering leg preference.
Physiscal Therapy in Sport, 11(4), 136–142. doi:10.1016/j.
ptsp.2010.06.005
Carpes, F. P., Rossato, M., Faria, I. E., & Mota, C. B. (2007a). Bilateral
pedalling asymmetry during a simulated 40km cycling time-trial.
Journal of Sports Medicine and Physical Fitness, 47(1), 51–57.
Carpes, F. P., Rossato, M., Faria, I. E., & Mota, C. B. (2007b). Influence of
exercise intensity on bilateral pedaling symmetry. In M. Duarte & G.L.
Almeida (Eds.), Progress in Motor Control IV (11th ed., Vol. 11, pp. s54-
55). São Paulo: Human Kinetics.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd
ed., pp. xxi, 567). Hillsdale, NJ: L. Erlbaum Associates.
Crenshaw, S. J., & Richards, J. G. (2006). A method for analyzing joint
symmetry and normalcy, with an application to analyzing gait. Gait &
Posture, 24(4), 515–521. doi:10.1016/j.gaitpost.2005.12.002
Daly, D. J., & Cavanagh, P. R. (1976). Asymmetry in bicycle ergometer
pedalling. Medicine and Science in Sports, 8(3), 204–208.
De Ruiter, C. J., De Korte, A., Schreven, S., & De Haan, A. (2010). Leg
dominancy in relation to fast isometric torque production and squat
jump height. European Journal of Applied Physiology, 108(2), 247–255.
doi:10.1007/s00421-009-1209-0
Dingwell, J. B., Joubert, J. E., Diefenthaeler, F., & Trinity, J. D. (2008).
Changes in muscle activity and kinematics of highly trained cyclists
during fatigue. IEEE Transactions on Biomedical Engineering, 55(11),
2666–2674. doi:10.1109/TBME.2008.2001130
Dunlap, W. (1999). A program to compute McGraw and Wong's common
language effect size indicator. Behavior Research Methods, Instruments &
Computers, 31(4), 706-709. doi:10.3758/BF03200750
Edeline, O., Polin, D., Tourny-Chollet, C., & Weber, J. (2004). Effect of
workload on bilateral pedalling kinematics in non-trained cyclists.
Journal of Human Movement Studies, 46(6), 493–518.
Fang, Y., Fitzhugh, E. C., Crouter, S. E., Gardner, J. K., & Zhang, S. (2016).
Effects of workloads and cadences on frontal plane knee biomechanics
in cycling. Medicine and Science in Sports and Exercise, 48(2), 260–266.
doi:10.1249/MSS.0000000000000759
Fohanno, V., Nordez, A., Smith, R., & Colloud, F. (2015). Asymmetry in elite
rowers: Effect of ergometer design and stroke rate. Sports Biomechanics,
14(3), 310–322. doi:10.1080/14763141.2015.1060252
Garcia-Lopez, J., Diez-Leal, S., Larrazabal, J., & Ogueta-Alday, A. (2015). No
bilateral asymmetry during pedalling in healthy cyclists of different per-
formance levels. 33rd. Poitiers, France: ISBS.
Gates, D. H., & Dingwell, J. B. (2008). The effects of neuromuscular fatigue on
task performance during repetitive goal-directed movements.
Experimental Brain Research, 187, 573–585. doi:10.1007/s00221-008-1326-8
Gouwanda, D., & Senanayake, A. (2011). Identifying gait asymmetry using
gyroscopes – A cross-correlation and normalized symmetry index approach.
Journal of Biomechanics, 44(5), 972–978. doi:10.1016/j.jbiomech.2010.12.013
Gregersen, C. S., & Hull, M. L. (2003). Non-driving intersegmental knee
moments in cycling computed using a model that includes three-
dimensional kinematics of the shank/foot and the effect of simplifying
assumptions. Journal of Biomechanics, 36(6), 803–813. doi:10.1016/
S0021-9290(03)00014-9
Haddad, J. M., Van Emmerik, R. E. A., Whittlesey, S. N., & Hamill, J. (2006).
Adaptations in interlimb and intralimb coordination to asymmetrical
loading in human walking. Gait & Posture, 23(4), 429–434. doi:10.1016/j.
gaitpost.2005.05.006
Heiderscheit, B. C., Hamill, J., & Van Emmerik, E. A. (2002). Variability of
stride characteristics and joint coordination among individuals with
unilateral patellofemoral pain. Journal of Applied Biomechanics, 18,
110–121. doi:10.1123/jab.18.2.110
Hentschke, H., & Stüttgen, M. C. (2011). Computation of measures of effect
size for neuroscience data sets. The European Journal of Neuroscience,
34(12), 1887–1894. doi:10.1111/j.1460-9568.2011.07902.x
Herzog, W., Nigg, B. M., Read, L. J., & Olsson, E. (1989). Asymmetries in
ground reaction force patterns in normal human gait. Medicine Science
in Sports and Exercise, 21(1), 110–114. doi:10.1249/00005768-
198902000-00020
Hreljac, A., & Ferber, R. (2006). A biomechanical perspective of predicting
injury risk in running. International Sportmed Journal, 7(2), 98–108.
Hug, F., & Dorel, S. (2009). Electromyographic analysis of pedaling: A
review. Journal of Electromyography and Kinesiology, 19(2), 182–198.
doi:10.1016/j.jelekin.2007.10.010
Lathrop-Lambach, R. L., Asay, J. L., Jamison, S. T., Pan, X., Schmitt, L. C.,
Blazek, K., & Chaudhari, A. M. W. (2014). Evidence for joint moment
asymmetry in healthy populations during gait. Gait & Posture, 40(4),
526–531. doi:10.1016/j.gaitpost.2014.06.010
Li, L., & Caldwell, G. (1999). Coefficient of cross correlation and the time
domain correspondence. Journal of Electromyography and Kinesiology, 9
(6), 385–389. doi:10.1016/S1050-6411(99)00012-7
Nigg, S., Vienneau, J., Maurer, C., & Nigg, B. M. (2013). Development of a
symmetry index using discrete variables. Gait & Posture, 38(1), 115–119.
doi:10.1016/j.gaitpost.2012.10.024

Nikodelis, T., Kollias, I., & Hatzitaki, V. (2005). Bilateral inter-arm coordina-
tion in freestyle swimming: Effect of skill level and swimming speed.
Journal of Sports Sciences, 23(7), 737–745. doi:10.1080/
02640410400021955
Pappas, E., & Carpes, F. P. (2012). Lower extremity kinematic asymmetry in
male and female athletes performing jump-landing tasks. Journal of
Science and Medicine in Sport, 15(1), 87–92. doi:10.1016/j.
jsams.2011.07.008
Park, K., Dankowicz, H., & Hsiao-Wecksler, E. (2012). Characterization of
spatiotemporally complex gait patterns using cross-correlation sig-
natures. Gait & Posture, 36(1), 120–126. doi:10.1016/j.
gaitpost.2012.01.016
Patterson, K. K., William, H. G., Brooks, D., Black, S. E., & McIlroy, W. E.
(2010). Evaluation of gait symmetry after stroke: A comparison of
current methods and recommendations for standardization. Gait &
Posture, 31(2), 241–246. doi:10.1016/j.gaitpost.2009.10.014
Pohl, M. B., Messenger, N., & Buckley, J. G. (2006). Changes in foot lower
limb coupling due to symmetric variations in step width. Clinical
Biomechanics, 25(6), 552–556.
Sanderson, D. J., & Black, A. (2003). The effect of prolonged cycling on
pedal forces. Journal of Sports Sciences, 21(3), 191–199. doi:10.1080/
0264041031000071010
Sayers, M., Tweddle, A., Every, J., & Wiegand, A. (2012). Changes in drive phase
lower limb kinematics during a 60min cycling time trial. Journal of Science
and Medicine in Sport, 15(2), 169–174. doi:10.1016/j.jsams.2011.09.002
JOURNAL OF SPORTS SCIENCES 2163
Smak, W., Neptune, R. R., & Hull., M. L. (1999). The influence of pedalling
rate on bilateral asymmetry in cycling. Journal of Biomechanics, 32(9),
899–906. doi:10.1016/S0021-9290(99)00090-1
Sugiyama, T., Kameda, M., Kageyama, M., Kiba, K., Kanehisa, H., & Maeda, A.
(2014). Asymmetry between the dominant and non-dominant legs in
the kinematics of the lower extremities during a running single leg
jump in collegiate basketball players. Journal of Sports Science and
Medicine, 13(4), 951–957.
Umberger, B., & Martin, P. (2001). Testing the planar assumption during
ergometer cycling. Journal of Applied Biomechanics, 17, 55–62.
doi:10.1123/jab.17.1.55
Wu, G., & Cavanagh, P. (1995). ISB Recommendations for standardization in
the reporting of kinematic data. Journal of Biomechanics, 28(10), 1257–
1261. doi:10.1016/0021-9290(95)00017-C
Wu, G., Siegler, S., Allard, P., Kirtley, C., Leardini, A., Rosenbaum, D., &
Stokes, I. (2002). ISB recommendation on definitions of joint coordinate
system of various joints for the reporting of human joint motion—Part
I: Ankle, hip, and spine. Journal of Biomechanics, 35(4), 543–548.
doi:10.1016/S0021-9290(01)00222-6
Yu, B., Gabriel, D., Noble, L., & An, K. N. (1999). Estimate the optimum
cutoff frequency for the butterworth low-pass digital filter. Journal of
Applied Biomechanics, 15(3), 319–329. doi:10.1123/jab.15.3.318
Zifchock, R. A., Davis, I., Higginson, J., & Royer, T. (2008). The symmetry
angle: A novel, robust method of quantifying asymmetry. Gait &
Posture, 27(4), 622–627. doi:10.1016/j.gaitpost.2007.08.006
Copyright of Journal of Sports Sciences is the property of Taylor & Francis Ltd and its
content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email
articles for individual use.