ESMAC 2012 abstract / Gait & Posture 38 (2013) S1–S116 S113

better (2–7◦). The result indicates the necessity of standardizingmeasurements. Further analysis including more subjects will beperformed.

http://dx.doi.org/10.1016/j.gaitpost.2013.07.227

P88

Underwater Gait Analysis: A MarkerlessApproach

Alberto Rigato 1, Zimi Sawacha 1, SilviaFantozzi 2,3, Francesco Pretto 1, Alice Mantoan 1,Matteo Cortesi 3, Silvia Del Din 1, Giorgio Gatta 3,Claudio Cobelli 1

1 University of Padova, Department of InformationEngineering, Padova, Italy2 University of Bologna, DEIS, Bologna, Italy3 University of Bologna, Faculty of Exercise and SportSciences, Bologna, Italy

Introduction: The measurement of common biomechanicalparameters during water locomotion is more demanding than inlaboratory conditions, since most instruments are not suitablefor operating in a water environment. Therefore, the develop-ment of new technologies is highly sought. In the present work,an automatic markerless motion capture system (MMC) [1] hasbeen investigated and its accuracy in underwater (UW) three-dimensional (3D) lower limbs (LL) joint kinematics during walkingreconstruction has been tested.

Patients/Materials and Methods: Three healthy male subjectswere recruited (mean age and BMI: 33.3±15.7 and 24.1±3.2). Avideo-based, MMC for the UW motion analysis was adopted asin Ceseracciu et al. [1]. Six walking trials at a self-selected speedhave been acquired with 6 subaqueous video cameras (TS-6021PSC,Tracer Technology Co. Ltd) in a swimming pool with water at1.20 m. Lower limb 3D joints kinematics was determined for eachsubject’s walking trial. The Pearson’s correlation coefficient wascalculated for each subject’s kinematic parameter, in order to selectwhich of each subject’s representative walking trials were to beincluded in the computation of the mean. Walking trials with acorrelation coefficient lower than 0.75 were excluded from the sta-tistical analysis [2]. For each subject, each joint rotations mean±1SD were estimated. Inter- subjects mean and SD were calculatedand thus UW Markerless Bands were obtained and then comparedwith the state of the art, represented by Normative Out of Water(OW) Bands [2]. Accuracy and reliability of the proposed techniquewere evaluated by means of comparison with 3D joint kinemat-ics estimated with traditional manual digitization (MAN) using adedicated software (SIMI Reality Motion Systems GmbH). In orderto compare the 2 techniques, knee joint planar angles were evalu-ated on each subjects and root mean square distance (RMSD) valuesbetween angles estimated.

Results: Mean, SD and range of motion (ROM) of hip, knee andankle joints’ angles were reported in Table 1. RMSD mean valuebetween knee 2D angles evaluated with MAN and markerless tech-niques was found equal to 8.1◦.

Table 1Mean.SD and ROM of hip. knee and ankle joint angles.

Hip Knee Ankle

nean SD ROM Mean SD ROM Mean SD ROM

flex-ext[deg] -141 13.5 61.18 14.8 13.8 50.1 -14 6.0 37.0abd-add[deg] 3.4 3.0 20.0 - - - / / /inv-ev[deg] / / ./ / / / -17 3.0 7.6in-exrotation[deg] 14 0.4 15.6 - - - -06 32 11.1

Discussion & Conclusions: Results show the feasibility of thepresent approach. A limitation of the study is that, even thoughMMC has been conceived aiming at 3D analysis [1], in this contextplanar angles evaluation is required for the comparison with stateof the art [3]. Differences between the 2 methodologies are dueto the fact that SIMI planar angles estimation relies on anatom-ical landmarks identification on each video sequence by manualdigitization, meanwhile the MMC approach is based on automaticreconstruction of joint centre positions. Future developments willbe the extension of this study to a larger sample of subjects and thecomparison between 3D joint angles estimated with the 2 differenttechniques.

References

[1] Ceseracciu, et al. Markerless analysis of front crawl swimming. J of Biomechanics2011.

[2] Sawacha, et al. Clinical Biomechanics 2009;24(9):722–8.[3] Barela, et al. J Electromyogr Kinesiol 2008;18(3), 446- 54.

http://dx.doi.org/10.1016/j.gaitpost.2013.07.228

P90

Validation of a motion capture laboratory and anew marker placement protocol for clinicalapplications

Tim Weber 1, Silvia Dullien 1, Joachim Grifka 1,Tobias Renkawitz 1, Sebastian Dendorfer 2

1 University Medical Center, Department ofOrthopedics, Regensburg, Germany2 University of Applied Science, Laboratory forBiomechanics, Regensburg, Germany

Introduction: Clinical gait analysis has been proven to be a validtechnique in order to reveal pathologies that concern ones gait.The greatest potential lies in capturing even the smallest abnor-malities in ones gait. Body fixed sensors for gait analysis havebeen introduced several years ago. Following specific marker pro-tocols the (often) motion-capture (MoCap) markers are applied onanatomical (bony) landmarks on the subject’s body, which serve tocalculate joint kinematics. These results are of course dependenton the examiners themselves and on the used marker placementprotocol. This study introduces a method to measure the accuracyand sensitivity of a motion capture gait laboratory, a new markerplacement protocol and the examiners conducting the gait analysisin order make evident statements.

Patients/materials and methods: This study is divided intothree different experiments. First, the gait laboratory itself is com-pared to a state of art rigid body measuring device without humaninteraction. Ten markers are placed on a standard cross table inthe measuring volume of the gait laboratory. By means of the crosstable different spatial positions of the markers are achieved whilethe measuring devices are capturing the positions simultaneously.The positions are then compared amongst each other. Secondly, thetracking sensitivity of the markers is determined by means of oneexperiment on a healthy subject. This is being tracked ten times bythree different examiners and the positions of the MoCap – markersare compared for every time step, respectively. The third experi-ment was conducted in order to determine the difference and thesignificance of applying of the MoCap – markers on the bony land-marks of the subject. The marker protocol was applied ten timeson the bony landmarks of the subject by two different examinersin an alternating manner. After each application a gait experimentwas conducted, which resulted in comparing marker trajectoriesand joint kinematics.

S114 ESMAC 2012 abstract / Gait & Posture 38 (2013) S1–S116

Results: Differences between rigid body measuring device, theMoCap laboratory and the cross table were not severe enoughto limit the gait analysis capabilities of the system. The differ-ences between inexperienced and experienced examiners werepresent. The application of the marker placement protocol showeddifferences between the examiners, however once the marker tra-jectories are translated into joint kinematics the differences werenon-significant.

Discussion and conclusions: This study introduces a methodto measure the accuracy and sensitivity of a fully equipped motioncapture gait laboratory, a new marker placement protocol and theexaminers conducting the gait analysis. The results indicate differ-ences between examiners. However they mostly disappeared aftera learning curve of the persons. Also they had only little influenceon the kinematics derived from the marker trajectories. This arti-cle analyses a new marker protocol and gait lab set up on severallevels and thus validates the accuracy and sensitivity of the systemwhich is crucial in order to be able to make valid statements aboutgait kinematics.

http://dx.doi.org/10.1016/j.gaitpost.2013.07.229

P92

Ethnic specific 3D knee motion during daily-lifeactivities: Can Japanese and Caucasian referencedatabase be merged?

B. Callewaert 1, L. Scheys 2, J. Bellemans 3, S.Fukagawa 4, K. Desloovere 5

1 C-MAL, University Hospital, KU Leuven, Pellenberg,Belgium2 Smith & Nephew, European Centre for KneeResearch, Leuven, Belgium3 KU Leuven, Department of Orthopedics andTraumatology, Leuven, Belgium4 KKR Chihaya Hospital, Higashi-ku, Fukuoka-shi,Japan5 KU Leuven, Department of Rehabilitation Sciences,Leuven, Belgium

Introduction: Previous studies have suggested variations inknee joint biomechanics between Caucasian and Japanese popu-lations [1]. However very few studies have reported 3D kinematicsand kinetics during more complex daily-life motor tasks. Therefore,the objective of the study was to analyze 3D joint motion duringfunctional motor tasks in healthy Caucasian and Japanese subjects,with special focus on the knee joint.

Patients/materials and methods: Upon ethical approval, 40healthy adults participated in this study after giving informed con-sent. 20 Caucasian subjects (9 men, 11 women, 32.4 ± 9.3 years, BMI22.5 ± 3.5) and 20 Japanese subjects (10 men, 10 women, 30.1 ± 6.6years, BMI 20.7 ± 2) were included. Kinematic data were obtainedusing a 14 camera motion capture system tracking the 3D positionsof 23 retro-reflective markers (Plug-in-Gait, Vicon, Oxford). Elevenfunctional motor tasks, 3 repetitions each were performed by allsubjects, including gait tasks (walking, walk and crossover turn andsidestep turn, ascent onto a step, descent off a step, descent withcrossover turn and sidestep turn) and non-gait tasks (sit-to-stand,mild and maximum squat and lunge. In addition, each subject wasasked to maximally rotate the trunk around the vertical axis, whilekeeping the knees extended and the feet fixed to the ground, induc-ing maximal axial rotations in the knee (thereby reflecting the levelof joint laxity).

Results: Japanese subjects showed a significant increased peakknee flexion angle (21.2%) for maximum SQUAT. We thereby also

observed a significant higher excursion of the coronal and trans-verse knee kinematics with a significant increased peak internalrotation angle of the knee (54.0%). The mean knee adductionmoment during squat was higher in the Japanese group (160.6 N m)compared to the Caucasians (121.34 N m). Furthermore, trunk rota-tion was characterized by higher excursions for out-of-sagittalplane rotations (Table 1). For all GAIT tasks, a significant decreasedknee extension angle at terminal stance was found for the Japanesesubjects (2.4◦) compared to the Caucasians (6.4◦). The tendency toa more flexed knee gait pattern resulted in a premature internalknee extension moment at terminal stance. Gait speed was highercompared to the Caucasian subjects.

Discussion and conclusions: This study demonstrates the needfor an ethnic specific reference database for 3D knee motion dur-ing daily-life activities. Squatting is a common daily posture inAsian populations. The increased range in out-of-sagittal rotationsseen during trunk rotation confirms the suspected joint laxity,resulting in higher excursions during maximum squatting. Theincreased mean knee adduction moments induces higher loadson the knee joint. Several studies have demonstrated the asso-ciation of prolonged squatting and the incidence of tibiofemoralknee osteoarthritis [2]. Furthermore, despite the increased walk-ing speed, Japanese subjects showed decreased knee extensionin stance, compared to Caucasian populations, resulting in a pre-mature knee extension moment at terminal stance, and thussuggesting a slightly reduced energy conservation.

References

[1] Leszko, et al. Clinical Orthopaedics and Related Research 469(1):95–106.[2] Zhang, et al. Arthritis & Rheumatism 50(4):1187–92.

http://dx.doi.org/10.1016/j.gaitpost.2013.07.230

P94

EMG activity during robotic assisted gaittraining: A case study

Marianna Romei 1, Claudio Corbetta 2, LuigiPiccinini 2, Elena Beretta 3, Anna C. Turconi 2

1 IRCCS E. Medea, Bioengineering Department,Bosisio Parini (LC), Italy2 IRCCS E. Medea, Functional RehabilitationDepartment, Bosisio Parini (LC), Italy3 IRCCS E. Medea, Neurorehabilitation Department,Bosisio Parini (LC), Italy

Introduction: Robotic assisted gait training (RAGT) is a rehabil-itation strategy that promotes the recovery of walking followingneurological gait disorders, both in adult and in pediatric popula-tion [1,2]. Many training parameters can be set, among which thepercentage of body weight support, and the patient can performpassive or active movement. The lower limb muscle activity duringRAGT was investigated in adult subjects [3].

The aim of this case study was to compare EMG activity of lowerlimb muscles in a 10-year-old girl during RAGT according to differ-ent body weight supports and patient’s passive/active movement.

Patients/materials and methods: DL is a 10-year-old girlaffected with left hemiplegia due to vascular cerebral lesion, whohad RAGT on pediatric Lokomat (Hocoma, Zurich, CH).

During RAGT, bilateral surface EMG activity of soleus (SOL), vas-tus medialis (VAM) and biceps femoris caput longus (BFCL) wasrecorded using Free EMG (Bts, IT) at 1000 Hz for at least 20 s. Threetraining conditions were tested: A) 100% of body weight (BW) sup-ported by the Lokomat and passive movement; B) 100% of BW