See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/267809994

Three-dimensional kinematic analysis of shoulder through wearable inertial

and magnetic sensors during swimming strokes simulation

Conference Paper · January 2013

CITATIONS READS

0 221

5 authors, including:

Fabrício Magalhães Andrea Giovanardi

Federal University of Minas Gerais Adidas
36 PUBLICATIONS 121 CITATIONS 23 PUBLICATIONS 329 CITATIONS

SEE PROFILE SEE PROFILE

Matteo Cortesi Giorgio Gatta

University of Bologna University of Bologna
49 PUBLICATIONS 323 CITATIONS 64 PUBLICATIONS 567 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Motion analysis using inertial and magnetic sensors View project

Special Issue "Aquatic Experience in Physical Literacy: From Specific to Extensive Meanings of Aquatics for Life" View project

All content following this page was uploaded by Fabrício Magalhães on 05 November 2015.

The user has requested enhancement of the downloaded file.

 THREE-DIMENSIONAL KINEMATIC ANALYSIS OF SHOULDER THROUGH WEARABLE INERTIAL AND
MAGNETIC SENSORS DURING SWIMMING STROKES SIMULATION
1
Fabrício Anício Magalhães, 2Andrea Giovanardi, 2Matteo Cortesi, 2Giorgio Gatta and 1,2Silvia Fantozzi
1
Department of Electrical, Electronic and Information Engineering, University of Bologna, Bologna, Italy;
2
School of Pharmacy, Biotechnology, and Sport Science, University of Bologna, Bologna, Italy
Corresponding e-mail: [email protected]

SUMMARY Recently, several authors examined the difference of

Wearable inertial and magnetic measurement units movement patterns and the evaluation of the athlete's
(IMMUs) have recently gained much attention in technique using sensors composed by accelerometers,
investigations approaching human kinematic analysis in gyroscopes and magnetometers, also known as IMMUs,
sports scenario. Running, tennis, baseball, snowboarding, covering a large range of disciplines, including: ambulatory
rowing, and swimming are some examples of sports already measurements, physical activity, gait analysis, and
approached. In swimming, many studies proposed several improvement of the athlete’s performance [1]. Thus, the use
methods to identify the temporal phases of a single of IMMUs has been presented as a useful tool for
swimming stroke or the swimming style. However, this monitoring human movement kinematics.
temporal information alone does not provide any joint
In swimming, wearable IMMUs were used to measure
kinematic data like joint angles. Therefore, the present
several variables regarding athletes’ performance including
study aimed to verify the accuracy of a protocol, previously
lap time, stroke identification, stroke count, stroke rate,
developed for ambulatory joint kinematic analysis through
stroke length, wall push off, forward speed, swimmer's
IMMUs, in measure the shoulder kinematics during
proficiency, energy expenditure, and swimming velocity [2,
swimming strokes simulation. A stereo-photogrammetric
3]. However, to the knowledge of the present authors, no
system was considered the gold standard. Three trained
previous investigation performed a three-dimensional
swimmers realized 3 trials of breaststrokes simulation and 3
kinematic analysis of the shoulder joint, that is, by far, the
trials of front-crawl strokes simulation in dry condition. As
most demanded joint during swimming [4]. As a
first verification, the relative motion of both the segments
consequence, investigations approaching the shoulder
thorax and arm with respect to the first-synchronized frame
kinematics analysis can aid coaches and therapists in
(automatically detected by an ad-hoc algorithm) was
identifying risk factors for injuries as well as in planning
compared by means of root mean square error (RMSE) and
injuries prevention programs. Therefore, the aim of this
correlation coefficient (r) between the two systems. The
study was to verify the suitability and accuracy of a
RMSE was 5° and 7°, and the r was 0.85 and 0.91 for
protocol previously developed for the shoulder joint
breaststroke and front-crawl stroke, respectively, indicating
kinematic analysis through IMMUs in ambulatory settings
a good relationship between both methods in measuring the
in measure the shoulder joint kinematics during swimming.
body segments’ orientation. As second verification, the
shoulder flexion/extension, abduction/adduction and METHODS
internal/external rotation angles were computed, and no The protocol implemented was described and validated by
significant difference was found (p<0.05) between both Cutti and co-workers [5]. Whereas the protocol was
systems. In conclusion, a protocol previously implemented developed to measure the upper-limbs kinematics in
for joint measurement in ambulatory settings is also ambulatory settings, the present work proposes to verify its
suitable and accurate to estimate the shoulder kinematics applicability also in swimming.
during swimming strokes simulation when using wearable
inertial and magnetic measurement units. From a biomechanical point of view, the shoulder joint was
modeled as an open kinematic chain composed by 2 rigid
INTRODUCTION segments (thorax and arm), with 3 degrees of freedom. Two
The use of technology in swimming played an important clusters composed of one IMMU and four retro-reflexive
role in acquiring reliable performance data to provide passive markers fixed on a wooden plate (15x15x1cm)
greater understanding of the swimming biomechanics and were used, one placed on the right arm and the other on the
enable swimmers to perform to their highest potential. thorax. Basically the protocol consisted in positioning the
clusters on the body segments and in calculating joint
angles according to the appropriate Euler’s convections.
The evaluation of the IMMUs’ estimation of the orientation its percentage of the duration were plotted overlapped. No
was performed during swimming strokes simulation on a significant difference was found (p<0.05) between both
bench in dry condition. A wireless IMMUs system (APDM, systems in the estimation of the shoulder flexion/extension,
Opal, USA, 2 nodes, 128Hz) and a stereo-photogrammetric abduction/adduction and internal/external rotation angles.
system (BTS Smart DX, Italy, 8 cameras, 200Hz) recorded Nonetheless, these values are comparable to the three-
the trials. Data from both systems were posteriorly dimensional joint kinematics estimated using underwater
synchronized and resampled at the IMMUs’ frequency rate. kinematics video analysis [6].
Three trained swimmers were laid facing down on a bench
For our analysis, only the shoulder was considered because
and their lower-limbs were hold tight by a person. Each
it is the most demanded joint during swimming. Further
subject performed 2 trials (one breaststroke and one front-
investigations in other joints such as elbow, wrist, hip, knee
crawl stroke) during 10s, trying to simulate the movements
and ankle can be addressed in order to get a full body
in the swimming pool. The relative motion of the segments
kinematic analysis during swimming.
thorax and arm with respect to the first-synchronized frame
was compared by means of root mean square error (RMSE) In this work we intended to perform shoulder kinematic
and correlation coefficient (r) between the two methods. analysis in dry condition because the protocol proposed by
Cutti and co-workers [5] was implemented in an
RESULTS AND DISCUSSION
ambulatory environment. The dry condition has two main
The goal of the present study was to verify the accuracy of
advantages: 1) the stereo-photogrammetric system is more
a protocol previously developed for ambulatory joint
accurate with respect to the underwater one, and 2) the
kinematic analysis through IMMUs in measure the shoulder
whole swimming stroke cycle can be analyzed: the aerial
kinematics during swimming strokes simulation in
phase (recovery) as well as the “underwater” phases.
comparison with a gold standard system. The results of the
Finally, the joint kinematic analysis in real condition (i.e.
first verification are presented in table 1.
underwater) during swimming will be aimed in the next
Table 1: Root mean square error and correlation coefficient future.
for both measurement systems. CONCLUSIONS
Breaststroke Front-Crawl The protocol implemented previously for joint
RMSE (°) r RMSE (°) r measurement in ambulatory settings is also suitable and
Mean 5 0,85 7 0.91 accurate to estimate the shoulder kinematics during
Minimum 3 0.76 5 0.82 swimming strokes simulation using wearable inertial and
Maximum 8 0,97 10 0.97 magnetic measurement units.
REFERENCES
The mean RMSE was 5° and 7° for the breaststroke trials 1. Callaway, A.J. et al., International Journal of
and for the front-crawl stroke trials, respectively, in Sports Science & Coaching. 4(1):139-153, 2009.
accordance with other investigations that performed human 2. Dadashi, F., et al., Sensors. 12(10):12927-12939,
joint angle measurement. In addition, the mean r was 0.85 2012.
for the breaststroke trials and 0.91 for the front-crawl stroke 3. Ohgi, Y. Proceedings of IEEE on Sensors, 2002.
trials indicating a good relationship between both systems 4. Heinlein, S.A. and Cosgarea, A.J. Sports Health,.
in estimating the body segments’ orientation with respect to 2(6):519-525, 2010.
the first frame. 5. Cutti, A.G., et al., Medical & Biological
Figure 1 shows the three shoulder angles for one subject’s Engineering & Computing. 46(2):169-178, 2008.
trial representative of all trials. Five strokes normalized by 6. Ceccon, S. et al., Journal of Sports Science. 1-12,
2012.

Figure 1: Shoulder flexion/extension, abduction/adduction and internal/external rotation. Lines green for the
stereo-photogrammetric system (BTS) and lines black for the inertial and magnetic measurement units (Opal).

View publication stats