[Paolo b. pascolo_(eds.)]_biomechanics_and_sports

CISM Courses and Lectures No. 473
BIOMECHANICS
ANDSPORTS
edited by PAOLO B. PASCOLO
ERRATA CORRIGE
XXI Winter Universiads 2003
instead of
XI Winter Universiads 2003

CISM COURSES AND LECTURES
Series Editors:
The Rectors
Manuel Garcia Velarde - Madrid
Mahir Sayir - Zurich
Wilhelm Schneider - Wien
The Secretary General
Bernhard Schrefler - Padua
Executive Editor
Carlo Tasso - Udine
The series presents lecture notes, monographs, edited works and
proceedings in the field of Mechanics, Engineering, Computer Science
and Applied Mathematics.
Purpose of the series is to make known in the international scientific
and technical community results obtained in some of the activities
organized by CISM, the International Centre for Mechanical Sciences.

INTERNATIONAL CENTRE FOR MECHANICAL SCIENCES
COURSES AND LECTURES - No. 473
BIOMECHANICS AND SPORTS
PROCEEDINGS OF THE
XI Winter Universiads 2003
EDITEDBY
PAOLO B. PASCOLO
UNIVERSITÂ. DI UDINE
" Springer-Verlag Wien GmbH

This volume contains 125 illustrations
This work is subject to copyright.
AlI rights are reserved,
whether the whole or part of the material is concemed
specifically those of translation, reprinting, re-use of illustrations,
broadcasting, reproduction by photocopying machine
or similar means, and storage in data banks.
© 2004 by Springer-Verlag Wien
Originally published by CISM, Udine in 2004.
SPIN 10992210
In order to make this volume available as economicalIy and as
rapidly as possible the authors' typescripts have been
reproduced in their original forms. This method unfortunately
has its typographicallimitations but it is hoped that they in no
way distract the reader.
ISBN 978-3-211-21210-3 ISBN 978-3-7091-2760-5 (eBook)
DOI 10.1007/978-3-7091-2760-5

PREFACE
On XII Winter Universiads 2003, CISM offered its scientific contribution by hosting a
confererence on mechanics applied to sports and, in general, to human movement.
A systematic debate on few specialized topics was out of the scope ofthe meeting;
rather, the conference was conceived as a chance to overview experiences gainedfrom
several operators working on dijferent aspects of biomechanics. Furthermore, not
aiming at a comprehensive cover ofsuch a complex argument, only some topics have
been dealt with during the conference.
In this way the reader will face in these proceedings bioengineering aspects,
control issues, techniques for the optimization of human performances as well as
methods for the improvement ofathletic equipments and devices. Biomechanical data
and signal processing, biomaterials and robotics complete the proposedframework.
Further works were included in the poster session of the conference and are not
presented here. We just mention an innovative use of a multibody code (Adams by
MscSoftware) for the improvement ofthe design ofski-boots and some investigations
on paraplegic subjects regarding electro-stimulated pedalling and optimisation ofthe
wheel-chair propulsion.
Some works were consistent with the fact that 2003 was designated as European
Year ofDisabled People. Indeed, many innovations in sport and biomechanics could
suggest interesting rehabilitative applications and a better prevention of some
pathologies due to the exercise ofsome normal activities like professional cycling.
We hope that, even in the future, sport events like Universiads could be associated
to scientijic initiatives like the one presented here.
Paolo B. Pascolo

CONTENTS
Preface
by P. B. Pascolo
The Biomedical Engineering Education in Italy
by M Bracale ................................................................................................................ 1
Video-Fluoroscopy Based Investigation of Intervertebral
Kinematics for Sport Medicine Application
by M Sansone, P. Bifulco, M Cesarelli and M Bracale ..............................................5
Computation of Rigid Body Motion Parameters from Video-
Based Measurements
by U. Tarantino, D. Perugia, G. Campanacei and E. Pennestri .................................. 11
Mechanical Ventilators and Ventilator Testers
by G. Belforte, G. Eula and T. Raparelli ......................................................................27
Cardiovascular and Metabolic Effort in a World Class Sailor
at Different Wind Velocities
by T. Prinei, C. Capelli, G. Delbel/o andL. Nevierov..................................................37
A mechanica1 model of the biceps brachii muscle
by M Gatti ,P. Pascolo, N. Rovere and M Saccavini ...............................................43
Evaluation of Quadriceps Muscles in Anterior Knee Pain: a
Possible Sport Medicine Application
by M Cesarel/i, P. Bifulco, M Sansone, M Romano and
M Bracale .................................................................................................................... 53
A Neural-based Model for the Control of the Arm During
Planar Ballistic Movements
by S. Con/orto, M Schmid, G. Gal/o, T. D 'Alessio,
N. Accornero and M Capozza......................................................................................59
The Relevance of Auditory Information in Optimizing Hammer
Throwers Performance
by T. Agostini, G. Righi, A. Galmonte, and P. Bruno..................................................67

Complex Test of Cycling Performance
by Z. Knol/, L. Kocsis, 1. Gy6re and R. Kiss...............................................................75
Foot-Floor Interaction in Classic Dancers
by C. Giacomozzi, S. Marucci, V. Macel/ari, L. Uccioli and
E. D 'Ambrogi..............................................................................................................89
Gait Pattern ofProfessional Fencers
by Z. Knol/, L. Kocsis andR. Kiss ..............................................................................97
Gait Alterations on Carriers of Bilateral Arthroplasty ofthe Hip
Suffering from LES: Clinical, Radiographic and Instrumental
Evaluation with Gait Analysis
by M Bacchini, C. Rovacchi and M Rossi............................................................... 111
Quantification with Gait Analysis of Biomechanic Risk
Protofactors Regarding the Patellar Tendinosis in Athletes with
Varus Knee
by M Bacchini and M Rossi.................................................................................... 123
Teaching a Robot with Human Natural Movements
by G. Magenes and E. Secco .................................................................................... 135
Numerical Simulation ofMotorcycles Crash Test
by L. Fabbri, G. Franceschini and F. Mastrandrea .................................................147
Biomechanical Power Analysis in Nordic and Alpine Skiing
by A. Schwirtz, D. Hahn, A. Huber, A. Neubert andF. Tusker ................................ 161
3-D Kinematic and Kinetic Analysis ofG-Slalom at Valbadia
Cup-Race in 2002
By R. Pozzo, A. Canclini, C. Cotelli and G. Baroni ................................................. 169

The Biomedical Engineering Education in Italy
M. Bracale
Department of Electronic Engineering and Telecommunications - Biomedical Engineering Unit
University 'Federico II' of Naples, Italy
Abstract Biomedical Engineering teaching activities in Italy started in the years
1968-69. In 2000, about 150 undergraduate courses in Biomedical Engineering were
active at 19 Italian Universities, while PhD Courses in Bioengineering and Post-
graduate courses were available in 9 Universities. Accordingly to the reform of
the European Higher Education, since 2001, a new educational path was adopted
in Italy, consisting of a first level degree (3 years) eventually followed by a second
level, specialist degree (2 years) and lor Masters and finally by the PhD. At moment,
the Universities of Ancona, Bologna, Genova, Milano, Napoli, Padova, Pavia, Pisa,
Roma 1 'la Sapienza', Roma Campus Biomedico, Torino and Trieste offer the first
level degree (3 years) in Biomedical Engineering and in the next future will offer
the specialist degree (2 years). Other Universities offer specific cur-ricula or some
modules of Biomedical Engineering within other engineering degree courses.
1 Introduction
This paper presents the scenario of the Biomedical Engineering educat'lon in Italy. Biomed-
ical engineering teaching activities in Italy started in the years 1968-69. The course of
'Biomedical Electronics' was activated at the Univ. of Padova and at the Univ. of Naples
and the courses of 'Bio-energetic' and 'Biological Electronics' started at the Poly-technic
of Milan. Since then many other courses and educational programmes have started in
many Italian University. Post-graduate courses started in the years 1971-72 when 'Post-
graduated courses of Biomedical instrumentation' was set-up at the Univ. of Naples,
while Bioengineering PhD courses were activated in 1982 as consortium of various Uni-
versities with two administrative headquarters in Milan and Bologna. In 2000, about 150
undergraduate courses in Biomedical Engineering were active at 19 Italian Universities
(Ancona, Bologna, Brescia, Firenze, Genova, Milano, Modena e Reggio Emilia, Napoli,
Padova, Pavia, Pisa, Roma 1 'la Sapienza', Roma 2 'Tor Vergata', Roma 3, Roma Cam-
pus Biomedico, Sassari, Siena, Torino and Trieste), while PhD and postgraduate courses
in Bioengineering were available in 9 Universities (Bracale, 2002, Biondi and Cobelli,
2001, A.I.I.M.B. website).
Accordingly to the reform of the European Higher Education, since 2001, a new
educational path was adopted in Italy, consisting of a first level degree (3 years) eventually
followed by a second level, specialist degree (2 years) and lor Masters and finally by the
PhD. At moment, the Univ. of Ancona, Bologna, Genova, Milano, Napoli, Padova,
Pavia, Pisa, Roma 1 'la Sapienza', Roma Campus Biomedico and Torino offer the first

2 M. Bracale
level degree in Biomedical Engineering and in the next future will offer the specialist
degree. Other Universities offer specific modules of Biomedical Engineering within other
engineering degree courses.
2 Methods and Materials
University degree programmes in Biomedical Engineering at both undergraduate and
postgraduate level are provided in Italy. Accordingly with the new definitions of the Ital-
ian Ministry of Education, Universities and Research, there are two Scientific-Disciplinary
Sectors (Le. homogeneous scientific-educational topics or areas) concerning Bio-medical
Engineering education at University level: 'Electronics and Informatics Bioengineering'
ING-INF/06 and 'Industrial Bioengineering' ING-IND/34. At present, there are a total
of 83 teachers (of which 27 full prof., 28 associate prof. and 28 researchers) belonging to
ING-INF/06 sector and there are a total of 24 persons (of which 9 full prof., 11 associate
prof. and 4 researchers) belonging to ING-IND/34 sector.
In Italy the PhD in Bioengineering is a research degree, usually of 3 years duration.
Entry to Doctoral study is al-Iowed to postgraduates (until now, students who took a 5
years university degree), where having a Master's degree or a post-graduate specialisation
is not a prerequisite.
Since 1982, the Italian scientific community of Bioengineering annually organises
monothematic schools held in Bressanone of a duration of few days about specific Biomed-
ical Engineering topics. Every year many students (undergraduate and postgraduate)
and teachers participate to the school.
In Italy engineers who intend practice the engineering profession have to register with
the Italian Council of the En-gineers (Ordine degli Ingegneri).
Recently, after the new reform of university education, the Council of the Engineers
decided to form two sections. Section A for those having the Specialised Degree in
Engineering (giving the professional title of Engineer) and Section B for those having
only the Degree in Engineering (giving the professional title of Junior Engineer).
3 Results
At present, biomedical engineers in Italy usually do not undergo additional training to
their education.
On the contrary of Medical Physicists, in Italy there is not yet any recognition of
the Clinical Engineer by the Na-tional Health Service. Eventual training for engineers
employed in hospital, healthcare structures or industries is op-tional and, in general,
carried out independently. At present, in Italy there is not yet an accreditation process
for the University education nether for specific training. After the ministerial decree
(DM 509/99) on autonomy in the sphere of teaching, some of the traditional customs
and habits of Italian universities are changing, passing from the traditional system of
one level (the degree) to a system organised around different levels (the degree, the
specialised degree, the master's degree, and the doctorate of research), aligned with
the goal of a European space of higher education which is contained in the 'Bologna
Declaration'. Recent documents already contain the new general organisational principles

The Biomedical Engineering Education in Italy 3
and present the characteristics that the university system should have acquired at the
end of the process of innovation, among which there is the accrediting of the courses
of study (a system of certification based on the cultural value of a qualification derived
from university studies, on the meeting of the social and economic demand, and on the
suitability of the resources deployed by universities).
In particular, on July 2001, the National Committee for the Evaluation of the Uni-
versity System (Italian Ministry for Education Universities and Research), issued a doc-
ument (Doc 12/01) about the 'Activation of an accrediting system of the degree courses
in the Italian Universities: first recommendation and proposals'
Furthermore, the Institution 'National System for the Accreditation of the Courses
of Study in Engineering' (S.I.N.A.I.) will be soon constituted. The aim of the SINAI
Institution is to determine the Minimal RequireJ;nents (in terms of Credits and arguments
covered) in order that a Courses of Study would be 'accredited' (i.e. to get recognition
at national and eventualIy international level). At moment the Accreditation matter
is to the attention of the CRUI (The Italian Rectors' Conference) and of the National
Committee for the Evaluation of the University System.
Concerning the Biomedical Engineering Education, on April 2001 the Education Com-
mittee of the National Group of Bioengineering submitted to the Committee of the Deans
a proposal about the arguments of the courses of Biomedical engineering. On July 2001,
the Committee of the Deans transmitted a new proposal to receive comments, remarks
and objections in order to prepare soon a document about the Minimal Requirements
for the Courses of Study in Engineering.
4 Discussion and Conclusion
The 2000-2001 academic year witnesses a historic transition because Italy will pass from
the traditional system of one level (the degree) to a system organised around differ-
ent levels (the degree, the specialised degree, the master's degree, and the doctorate of
research), aligned with the goal of the Bologna declaration.
The credits system have been adopted to be in line with the ETCS European system
of credits, in which credits go from 1 to 60, are based upon the course unit, and describe
the total work burden which each course unit requires.
The challenge which universities are now facing is a colossal one, and it has forced
them to revise alI the university curricula and to create new ones adapted to a society
based upon knowledge which innovates and renews at extraordi-nary rhythms. In this
revision the universities are finalIy enjoying certain spaces of autonomy and in the interest
of the students are interacting with employers' associations and trade unions and other
state and private systems which are interested in university education and training.
Bibliography
A.I.I.M.B. http://www.aiimb.it.
E. Biondi and C. CobelIi. La jormazione dell'Ingegnere Biomedico. Patron editore, 2001.
M. Bracale. Biomedical engineering education in italy. In EAMBES documents, 2002.

Video-Fluoroscopy Based Investigation of
Intervertebral Kinematics for Sport Medicine
Application
M. Sansone, P. Bifulco, M. Cesarelli and M. Bracale *
Department of Electronic Engineering and Telecommunications - Biomedical Engineering Unit
University 'Federico II' of Naples, Italy
Abstract Spinal injuries can arise in many situations: on the road, at work, in
sport. Investigation of spine mechanics can be of help in the evaluation of the spinal
structures. Due to the natural inaccessibility and the complex structure of the spinal
segments, in vivo measurements of their mechanics are very problematic. Interver-
tebral kinematics is closely related to the state of the individual spinal segments
and then to spine functionality. Research on intervertebral motion has, therefore,
been widely regarded as an essential prerequisite to improve the knowledge of the
mechanics of the spine and its disorders. Clinical application of spine kinematics
analysis may include diagnostic assessment of spinal instability and evaluation of
surgical treatment. The aim of this script is to provide an historical perspective
about the methodologies for the analysis of spine motion, developed over the years
at our Department.
1 Introduction
Spinal functional alterations and related pathologies can generate various disabilities,
constituting a widespread problem, which continues to grow. Spinal injuries can arise
in many situations: on the road, at work, in sport. Investigation of spine kinematics
can be of help in the evaluation of the spinal structures.The mechanical functionality of
the spine depends on the dynamic behaviour of its components: the vertebrae, the discs
and the ligaments, in conjunction with the actions of the muscles. Due to the natural
inaccessibility and the complex structure of the spinal segments, in vivo measurements
of their mechanics are very problematic. Intervertebral kinematics is closely related to
the state of the individual spinal segments and then to spine functionality. Research on
intervertebral motion has, therefore, been widely regarded as an essential prerequisite
to improve the knowledge of the mechanics of the spine and its disorders. Possible
clinical application of spine kinematics analysis may include diagnostic assessment of
spinal instability and evaluation of surgical treatment. The aim of this script is to
provide an historical perspective about the methodologies for the analysis of spine motion,
*The authors wish to thank the private hospital 'Clinica Villalba of Prof. Umberto Bracale'
(Naples, Italy), with which the University of Naples has a non profit collaboration for scientific
purposes, for providing fiuoroscopic data and clinical support.

6 M. Sansone, P. Bifulco et al.
developed over the years at the Dept. of Electronic Engineering and Telecommunications
Biomedical Engineering Unit, University 'Federico II' of Naples, Italy.
2 Methods and .Nlaterials
Most of the analyses to characterise the intersegmental motion of the vertebral column
in vivo were carried out using plain radiography. These techniques were improved in
(Gianturco, 1944), measuring the angles between vertebral bodies at the extreme trunk
range in normal aud symptomatic subjects. More accurate and exhaustive kinematic
studies were carried out in vitro using cadaveric spinal segments. These studies provided
a 3D characterisation of the segmental motion of the lumbar (Rolander, 1966), thoracic
(White, 1969) and cervical (Lysell, 1969) spine.
Full 3D motion analysis in vivo has been attempted using biplanar radiographic
equipment (Brown et al., 1976). More reliable 3D kinematic intersegmental data can
be obtained in vivo by means of insertion of pins in the vertebrae (Steffen et al., 1997)
(generally utilised for pre-operative analysis, e.g. implantation of spinal fixators). The
possibility of using non-invasive methods such as flexible rules, inclinometers and go-
niometers or skin optical markers has been widely considered. However, these methods
are adequate for an entire section of the spine rather than for individual segments. Skin
and soft tissue effects also impede reliability of surface measurements. Most of the in
vivo studies employing conventional radiography, perform end-of-range measurements
(Dimnet et al., 1978). The number of exposures that can be obtained from one subject
(Cholewicki et al., 1991) is very limited to maintain radiation at an acceptable level and
only static images are produced.
From a diagnostic point of view, not only the extremes of movements, but also the
motion pattern in between, is of interest and can indicate underlying pathology (Pearcy,
1986). Recently, the use of digital video fluoroscopy has been proposed by different au-
thors to study in vivo intervertebral kinematics. This technique allows a more continuous
motion analysis, and can provide useful diagnostic data, maintaining radiation exposure
low enough to be acceptable for routine clinical application.
The use of a single fluoroscopic device limits analysis to planar motion of the spine.
This assumption is reasonable in some cases, as Pearcy and Bogduk (1988) reported
(see also Panjabi, 1979). Although flexion-extension movements generally occur without
significant lateral bending or axial rotation (i.e. coupled motion) this is not case for
lateral bending. However, even if, for the lateral bending of the lumbar tract, the amount
of coupled motion is relatively small with respect to the other tract, only the flexion-
extension in sagittal plane can be assumed to be a planar motion.
From a fluoroscopic sequence of images of the spine, a kinematic description of mo-
tion is based upon features of the vertebrae observed throughout the frames. A variety
of different features or landmarks (e.g. vertebral body edges or corners, processes or
pedicles) and measurement techniques have been proposed. For kinematic analysis the
hypothesis of rigidity must hold for the vertebrae. Such an assumption is natural since
deformation of the vertebrae caused by the forces acting on the vertebral column during
motion are negligible with respect to the displacement involved.
A range of kinematic indices have been reported in the literature to describe motion

Video-Fluoroscopy Based Investigation of Intervertebral Kinematics.....
.,. ---.., _"Y f
,
:-:·r,),Y
::.:t:r:c
(a)
Instrumental
set-up
(b)
Figure 1. (a) Instrumental set-up. (b) Software for automatic analysis
7
and among these are intervertebral angles, axis of motion, instantaneous centre of rota-
tion (ICR) and helical axis of motion. A biological significance has been proposed for
the ICRs a function of the centre of the reaction force of a vertebra. Accurate measure-
ments of the vertebrae positions throughout a motion sequence are required. This is due
to the relatively small range of motion of individual vertebral units and intrinsic errors
in the computation of the kinematic parameters (Panjabi, 1979). Nevertheless, manual
intervention (Van Mameren et al. , 1992) is stilliargely used for vertebrallandmark iden-
tification in spite of the fact that it is regarded as one of the major contributors to errors
(Panjabi, 1979). Moreover, the low X-ray dosage adopted for fluoroscopic analysis results
in poor quality image sequences which complicate the situation. This is particularly true
for the lumbar spine because of the larger amount of soft tissue involved.
3 Results
Recently, a method for automatic recognition of vertebral landmarks on fluoroscopic
images was proposed by Bifulco et al. (Bifulco et al. , 2001): this method was tested
using a calibration model giving good results in accuracy and precision. In Figure 1.a is
shown the instrumental set-up used for acquisition of fluoroscopic sequences; in Figure
l.b is shown the software for automatic analysis developed at our Deparment.
Anyway, most of the work previously described, was confined to the estimation of pla-
nar motion (mainly in the sagittal plane) and is based on the assumption of absence of
out-of-plane coupled motion (e.g. axial rotation). This assumption is reasonable for sagit-
tai (flexion-extension) movements (mainly due to anatomic symmetry), but it is certainly
erroneous for lateral bending where a coupled axial rotation is present (Gertzbein et al.,
1984). A knowledge of three-dimensional positioning of vertebrae against time could
lead to comprehensive 3D kinematic analysis, or at least an evaluation of the presence of

8
F
o o
t;
, .
t(
"~
.' ,..1.
f'. _ .. , ,
. ,
(a)
M. Sansone, P. Bifulco et al.
DRR
DRR

,, """"1 "" ""1/
, ",,(
, / 1,
""X
r """"""__ ,CTvolume
..... / --- "
Z '1 '
l'
Y I ' , '
..1: ''r'y x
(1))
Df?f?
--
Figure 2. (a) Digitally Reconstructed Radiograph. (b) 3D pose estimation
out-of-plane motion. The use of external reflecting skin-markers (Breen et aL, 1993), go-
niometers and other similar devices is appropriate for 3D gross-movement measurement
but not for intervertebral kinematics (due to skin-bone sliding). Some studies report
precise 3D intervertebral measurement by means of metallic pins inserted into vertebrac
(Cholewicki ct al., 1991), but such techniques are clearly inadequate for clinical applica-
tions. Other studies report the use of specialised apparatus such as stereo-radiography.
Alternatively, 3D vertebra positioning can be estimated by combining a single-plane
fluoroscopic projection with volumetric information provided by CT data, allowing easier
clinical application. Starting from CT data it is possible to digitally reconstruct radio-
graphic projections in different orientations thereby simulating the radiograph forma-
tion process (Bifulco et al., 2002). Comparing this Digitally Reconstructed Radiographs
(DRRs) with the fluoroscopic image it is possible to estimate the out-of-plane rotations
of a vertebra.
The 2D-3D registration is generally addressed by means of iterative algorithms, which
involve the optimisation of an appropriate cost-function. The cost-function which has
been proposed in (Bifulco et al., 2002), is the cross-correlation. A specific software for
automatic 3D pose estimation was developed at the Biomedical Engineering Unit, Dept.
of Electronic Engineering and Telecommunications (Bifulco et al., 2002). In Figure 2.a
is shown a typical method in computation of Digitally Reconstructed Radiograph. In
Figure 2.b is shown the approach 3D pose estimation of vertcbrae using cross-correlation
between DRRs and actual digital video fluoroscopic (DVF) images. A computer sim-
ulation (Bifulco et al., 2002) and an in vitro feasibility study (Sansone et al., 1999) of
the method has been performed using an animal vertebra rigidly fixed to a calibration

Video-Fluoroscopy Based Investigation of Intervertebral Kinematics..... 9
support. The results of the in vitro experiment were promising.
4 Discussion and Conclusion
From the above discussion emerges that further work is needed to develop the current
methodologies both for 2D and 3D motion analysis in order to make them more accurate,
precise, low invasive and possibly automatic: these are a key points to make the spine
kinematics reliable and routinely applicable as clinical examination. It is worthwhile to
mention that 3D pose estimation of generic skeletal structures within the fluoroscopic
field of view is can be of help also in intra-operative surgery, biomechanical evaluation of
prosthesis and radio-therapy planning.
Bibliography
P. Bifulco, M. Cesarelli, R. Allen, M. Sansone, and M. Bracale. Automatic recognition of
vertebral spine kinematics. Medical fj Biological Engineering fj Computing, 39:65-75,
200l.
P Bifulco, M Sansone, M Cesarelli, R Allen, and M Bracale. Estimation of out-of-plane
vertebra rotations on radiographic projections using ct data: a simulation study.
Medical Engineering fj Physics, 24:295-300, 2002.
A. C. Breen, R. Brydges, H. Nunn, J. Kause, and R. Allen. Quantitative analysis of
lumbar spine intersegmental motion. Eur. J. Physical Med. Rehab., 3:182-190, 1993.
B. Brown, A. Burnstein, C. Nash, and C. Schock. Spinal analysis using a three dimen-
sional radiographic technique. J. Biomech., 9:355-365, 1976.
J. Cholewicki, S. Mcgill, B. Wells, and H. Vernon. Method for measuring vertebral
kinematics from videofluoroscopy. Clin. Biomech., 6:73-78, 1991.
J. Dimnet, L. P. Fischer, G. Gonon, and J. P. Carret. Radiographic studies of lateral
flexion in the lumbar spine. J. Biomech., 11:143150, 1978.
S.D. Gertzbein, R. Holtby, M. Tie, A. Kapasouri, and B. Chan, K.W.and Cruickshank.
Determination of a locus of instantaneous centers of rotation of the lumbar disc by
moir fringes - a new technique. Spine, 9:409-413, 1984.
C. Gianturco. A roentgen analysis of the motion of the lower lumbar vertebrae in normal
individuals and in patient with low back pain. Am. J. Roentgend., 52:261, 1944.
E. Lysell. Motion in the cervical spine. Acta Orthop. Scand., 123, 1969.
M. Panjabi. Centers and angles of rotation of body joints: a study of errors and opti-
mization. J. Biomech., 12:911-920, 1979.
M. Pearcy. Measurement of back and spinal mobility. Clin. Biomech., 1:44-51, 1986.
M. Pearcy and N. Bogduk. Instantaneous axes of rotation of the lumbar intervertebral
joints. Spine, 13:1033-1041, 1988.
S. D. Rolander. Motion of the lumbar spine with special reference to the stabilizing effect
of posterior fusion. Acta Orthop. Scand., 90, 1966.
M Sansone, P Bifulco, M Cesarelli, and M Bracale. Estimation of the 3d positioning of
anatomic structures from radiographic projection and volume knowledge. In Proceed-
ings of EMBEC99, Vienna, pages 1005-1007, 1999.

10 M. Sansone, P. Bifulco et al.
T. Steffen, H. G. Rubin, R. K.and Baramki, J. Antoniou, D. Marchesi, and M. Aebi. A
new technique for measuring lumbar segmental motion in vivo. Spine, 22:156~166,
1997.
H. Van Mameren, H. Sanches, J. Beursgens, and J. Drukker. Cervical spine motion in the
sagittal plane (ii) position of segmental averagcd instantaneous centers of rotation~a
cineradiographic study. Spine, 17:467-474, 1992.
A. White. Analysis of the thoracic spine in man. Acta Orthop. Scand., 123, 1969.

Computation of Rigid Body Motion Parameters from
Video-Based Measurements
Umberto Tarantino *, and Dario Perugia *,
Giovanni Campanacci t and Ettore Pennestrl t
* Dipartimento di Chirurgia - Sez.Ortopedia
t Dipartimento di Ingegneria Meccanica
Universitâ di Roma Tor Vergata
Abstract The objective of the work is to compare and improve the accuracy of
existing methods for the computation of rigid body parameters from positions,
velocities and accelerations of a set of non colinear anatomicallandmarks. Instead of
the common Euler angles, Cardan angles or Bryant angles, the results are expressed
in terms of the screw axis parameters (i. e. axis versor and rotation angle). In clinical
analysis the physical meaning of this axis is surely better understood than the
named angles. The paper summarizes also the main steps of noteworthy algorithms
for the extraction of finite and infinitesimal screw motion parameters from noisy
measurements of markers positions, velocities and accelerations. The sensitivity to
data errors of the methods reviewed has been investigated by means of numerical
tests. A commercial human motion analysis was also used for the field testing.
1 Introduction
There are many applications where, from point measurements, the position, the velocity
and the acceleration of a body are required. For instance, in the field of biomechanics,
video telernetry is a common tool for gait or inverse dynamics analysis. In this last case,
the driving forces of upper limbs can be estimated by substituting the experimentally
measured kinematics into the equations of a dynamic model e.g.(Pennestrl et al., 2002).
Video telemetry methods usually return only the three dimensional position of a
discrete set of characteristic points (markeTs) attached to the body.
In the analysis of human motion, limbs movement is tracked by means cameras whose
number ranges from 2 to 8.
Then, the spatial coordinates of rnarkers attached to the limbs (bodies) are obtained
in two steps:
• digitalization of each recorded image;
• transformation of marker coordinates from the space of the camera image to an
inertial reference frame.
For spatial motion analysis, three is the minimum number of markers for each limb.
However, redundancy in the number of markcrs is usually recommended.

12 U. Tarantino, D. Perugia et al.
After data smoothing, velocity and acceleration components are computed through
numerical differentiantion.
The numerical values of marker cooniinates include errors from many sources (cali-
bration, skin elasticity, marker size, image resolution, etc.). Although the greatest care
is takell for bounding alI errors, the kinematic characteristics of the markers are not
numerically consistent with the hypothesis of rigid body motion.
Since kinematic and dynamic models are based on the hypothesis of rigid body motion
this is a serious drawback. Thus, there is the need to cornpute screw motion parameters
of each body from the markers position noisy measurements.
For this purpose most of the methods available in literature minimize the error of
motion parameters using the least squares optimality criterion.
The paper will review and compare some of these methods, then will propose an adap-
tative computational strategy for the estimate of rigid body motion screw parameters.
This strategy has been tested on the field and the main results will be reported in the
paper.
2 Finite screw motion
It is well knowu 1 that any rigid spatial finite motion can be reduced to a rotation about
an axis, represcnted by a versor {'U}, and a translation of module 6.so along the same
axis (see Figure 1). Iu matrix notation, this displacement is represented as follows
z ......... al
Screw motion axis
a
y
Figure 1. Finite rigid displacement
lThis property of rigid body motion is usually attributed to Chasles (Bottema and Roth, 1979).
However, the priority of discovery must be attributed to Mozzi del Garbo (176:3).

Computation of Rigid Body Motion Parameters from Video-Based Measurements 13
{
al x
} ~[D] {
ax
}aly a y
(2.1)
al z a z
1 1
where (Suh and Radcliffe, 1978; Pennestrl, 2001)
[D]= [[A] {so} - [A] {so} + ilso{u}
]O 1
[[A] {d} ] (2.2)
O 1 '
and [A] a 3 x 3 orthogonal matrix.
When expressed as a function of the versor {u} and rotation angle 'ljJ, thc rotation
matrix [A] takes the form2 :
[
u;V'ljJ + el/;
[A] = 1LxUyV'ljJ + ~zS'ljJ
UxUzV'ljJ - uyS'ljJ
u;V'ljJ +C'ljJuyuzV'ljJ - uxSt/J .
UX U yV7f;,,-uzS1j; uxuzV'ljJ+uyS1jJ 1
uyuzV'l/; + uxSt/J u;V'ljJ + C'ljJ
(2.3)
The components of {u} and the angle 'ljJ can be retricved from the elements aij
(i = 1,2,3, j = 1,2,3) ofthe rotation matrix as follows:
• Let
• Compute the rotation angle
'ljJ = 2 cos-l eo
• When 't/J # 27fn (n = 0,1,2, ... y, compute the cartesian components of {v,} =
{ux uy uz}T
a32 - a23
U x = 4eo sin ~ ,
a13 - a3l
u y = 1j; ,
4eo sin "2
a2l - a12
Uz = ---,
4eo sin ~
Reference Cheng and Gupta (1989) offers an interesting review of formulas used for
expressing spatial rigid displacements.
2For conciseness, we let V1/'o = 1- cos 1/'0, C1/'o = cos 1/'0 and S1/'o = sin 1/'0.
3The case of 1/'0 = 27fn is discussed in Pennestrl (2001).

3 Review of some methods for a finite screw motion
Let us dellote with {pt} anei {Ft} the vectors formed by the coordinates of the positions
of a point at time t and t + b..t, respectively. Then, for a rigid motion, we have
{Ft} = [A] {p;} + {d} (3.1)
For a set of n points of the same body, once defined the matrices
and
[P] = [pl P2 ... Pn]
the equation (3.1) generalizes iuto the following
[F] = [A] [p] +{d}{hf , (:3.2)
with {h}nxl = { 1 1 }T.
In the case of experimental point measurements, the coordinates of points 011 the
same body do not fulfill the rigidity conditioIl. Thus, equation (3.1) does not hold
exactly. However, it is useful to define a matrix [A] and a vector {d} such that
where
[P] ~ [A] [15] + {d}{h}T ,
lP] = [151 152
[P] = [P1 P2
Pn ] ,
Pn ] .
(3.:3)
(3.4)
(3.5)
reprcsent thc matriccs whose columns are formed by the experimentally measured coor-
dinates of points on the same body at time t and t + b..t, respectively.
The numerical definition of [A] and {d} obviously depend on the criterion used to
reach the best approximation.
In the following subsections the main steps of some relevant algorithms will be prc-
sented.
3.1 Method of Veldpaus aud others
F.E. Veldpaus published many papers on this topic, (Veldpaus et al., 1988; Spoor
and Veldpaus, 1980; Woltring et al., 1985; Heeren and Veldpaus, 1992), the following
algorithm has been summarized from reference Veldpaus et al. (1988).
1. Compute
1 n
{po} = - L {pt} ,
n
(3.6)
i=l
{Po} = ~ t {Pi} ,
n i=1
(3.7)
1~{- -} T
[G] = - L- Fi - Fo {pi - po} .
n
(3.8)
i=1

2. Compute [Gt adjoint4 matrix of [G].
3. Compute
gî = tr [Gf [G] ,
g~ = tr [[G]T [G]r
3
g3 = det [G]
4. Compute Pl and P2 by solving iteratively the following system of equations
5. Compute
pî - 2P2 = gî ,
P~ - 2plg3 = g~ .
[el = [Gf [G] + P2 [1] ,
[A] = ([Gt + Pl [G]) [C]-l
3.2 Method of Shiffiett and Laub
(3.9)
(3.10)
(3.11)
(3.12)
(3.13)
(3.14)
(3.15)
The following algorithm has been summarized from references Laub and Shiffiett
(1982) and Shiffiett and Laub (1995).
1.
[NI ~ [PI [[I] - {h}: {h} lIPIT (3.16)
2. Apply singular value decomposition to matrix [N]
[N] = [U] [~] [V] (3.17)
3.
[A] = [U] [V]T (3.18)
If det [A] = -1, then the matrix needs to be redefined as follows:
[A] = [UL [Vf (3.19)
where [UL = [Ul U2 -U3] and [U] = [Ul U2 U3].
4.
{d} = {h}; {h} ([P]- [A] [p]) {h} (3.20)
4For a 3 x 3 matrix [G] = [{gI} {g2} {g3} 1the following formula can be used
where the symbol - denotes the skew-symmetric matrix associated to a vector.

3.3 Gupta and Chutakanonta
This algorithm has been summarized from reference Gupta and Chutakanonta (1998).
In the subsequent formulas the matrices of markers coordinates are assumed of the
form:
1. Compute
with
UJ] = [~l ~2
[P] = [~l ~2
[~] = [~1l ~22 ~ ~ 1O O IT33 O
O O O IT44
(3.21)
(3.22)
(3.23)
(3.24)
2. Compute [~+] as follows: If ITj,j =1- O, then ITJ; = 1/ITjj else ITJ; = O, (j = 1,2,3,4).
3. Compute
[ Dn
Dl2 Dl3
1
[Al] = D21 D22 D23 (3.25)
D31 D32 D33
{d}T = { Dl4 D24 D34 } (3.26)
where
[D] = [P] [U] [~+] [V]T (3.27)
4. The matrix [Al] is an approximation of matrix [A], but can be refined in two
different ways.
• First type of refinement.
1. Execute the QR decomposition of [Al], such that [Al] = [RI] [UI].
2. Let [A] = [RI ]
• Second type of refinement.
1. Execute the SVD decomposition of [Al], such that [Al] = [Ur] [~r] [Vrf·
2. Let [A] = fUr] [Vrf
4 Theoretical comparison of methods for finite motion analysis
The criteria for the evaluation of the numerical performances are not unique e.g. (Gupta,
1997) and (Park, 1995). In this paper the matrices UJ] and [P] are generated by varying
the number of exact figures after the decimal point. For example, if only k decimal figures

are specified, then all the remaining 8 - k decimals in the point data are set equal to
zero5 .
It must be observed that in actual measurements, not necessarily the number of exact
decimal figures is constant. However, this approach should test the sensitivity of the
algorithms to the loss of precision due to unavoidable experimental errorsG• The algo-
rithms have heen tested also by introducing statistical errors with a gaussian distribution
Campanacci (2000).
For a screw motion characterized by the following data7 :
- Screw axis {ua } = {V; V; O}T;
- Rotation angle Iia = ~ rad;
- Translation vector {da} = {IlO }T .
Thus, when
equation (3.2) gives
[
2.612372436
[P] _ 0.387627564
- 1. 500000000
]
2.862372436
1.137627564
2.112372436
3.337117307
-0.337117307
3.724744872
9.036607051 10.9633929491
6.337117308 .
1
Once the values of {uc}, {Iic} and {de} are computed, the following error indices
are plotted in Figures 2, 3 and 4 as a function of the number of exact figures used in the
calculations.
The labels Guptal, Gupta2 and Gupta3 denote the results obtained using the algorithm
of Gupta and Chutakanonta (1998) with no refinement, first type of refinement and
second type of refinement, respectively.
5 AII computations were carried out in Fortran and in single precision.
6 As a rule of thumb, for video based human motion analysis systems, the accuracy of the
cartesian components of the markers is about 3/1000 the length of the calibration cube.
Thus, for a calibration cube with a side 1 meter long, the center of the marker is tracked with
an error of ~3 millimiters.
7 Subscripts a and c denote analytical and numerical values computed with single precision (i. e.
8 decimal figures).

18 u. Tarantino, D. Perugia et al.
10
0.1
~
;:::
0.01<1
1E-3
1E-4
1E-5
6.• ........,....••••••• -.6.
2 3
....••••• ..&.
....
•••• GLPta1
-~·GLPta2
-~. GLPta3
~Laub
~Veldpaus
.........
",-"';. ",- ,'. ",-
, ,- ,
, ', ', '.'
4 5
Number of exact decimal figures
Figure 2. D.u% .vs. number of exact decimal figures
IMSL was the mathematical library used for computing the singular value and QR
decompositions, as weB as other matrix operations. For the method of Veldpaus and
others, the authors used the Fortran subroutine DATDTM, retrieved from the web site
of the International Society of Biomechanics.
4.1 The adaptative refinement
There is the possibility to assign a weight fi to the i th marker (i = 1,2, ... , n).
The authors of this paper found that it was possible an improvement in accuracy over
the standard application of the described algorithms.
An adaptative strategy can be used for prescribing the values of li- For this pur-
pose, the algorithm of Veldpaus is iteratively applied. At the first step fi = 1 for
(i = 1,2, ... , n). At the second step, the values of the weights are assigned such that:
(4.1)
(4.2)

Computation of Rigid Body Motion Parameters from Video-Based Measurements
~o
~
<1
~
~
<1
10
0.1
0.01
1E-3
1E-4
10
0.1
0.01
1E-3
......... GLpta1
---I:r-. GLpta2
--4- GLpta3
---Laub
---Veldpaus
2 3 4
.................
Figure 3. 6.1/;% .VS. number of exact decimal figures
---Laub
--- Veldpaus
-A-GLpta
......~
5
1E-4~----~-----r----~----~----~~----~----~--~
2 3 4 5
Figure 4. 6.d% .vs. llumber of exact decimal figures
19

In other words the weight of the marker is proportional to the inverse of the estimated
position error. The same concept can be extended to the analysis of first and second order
instantaneous screw axes Campanacci (2000).
5 Experimental analysis with the adaptative refinement
The adaptative refinement has been tested on the field by analyzing the mohon of a 33
r.p.m. record player by means of the Veldpaus' algorithm without and with refinement.
5.1 The experimental setup
The motion of a record has been tracked by means of video based commercial teleme-
try system. The system made use of the APAS software, two 60 Hz Panasonic cameras
and one calibration cube with 1 m long edge. Five markers were attached to the record
Figure 5. View of markers and record player from the right camera
(see Figure 5).
Considered the position of the inertial reference system, (see Figure 6), the actual
transform matrix was
o O 1cos e sine ,
- sine cos e
(5.1)
with ethe angle of rotation of the disc. The measurement error has been defined as
suggested by Gupta Gupta (1997),
(5.2)
where
[~A] = [A,] - [A] . (5.3)
In Figure 7 the value of Ohas been plotted versus time.

Computation of Rigid Body Motion Parameters from Video-Based Measurements
1
,,,,
7 Y 6
.....---
8
.----,
4
5
Figure 6. Calibration cube and inertial cartesian system
6 Infinitesimal screw motion
Let us denote with
21
- {Vi} and {ad the velocity and acceleration vectors of the ith point, respectively;
- s, the speed of the points along the screw axis;
- {r} a vector of a generic point on the screw axis;
- {u} screw axis versor;
- {w} and {a} velocity and angular velocity vectors, respectively;
- {t} angular acceleration axis versor;
- {q} a vector of a generic point Q on the angular acceleration axis versor.
The velocity and the acceleration of a generic point on a rigid body can be expressed as
follows:
{vd = s{u} + [w] {Pi - r} ,
{ai} = {ao} + [ii] {Pi - q} + [02] {Pi - q} ,
where
6.1 Review of the method of Sommer
(6.1)
(6.2)
(6.3)
This subsection reports the main steps of the algorithm of Sommer Sommer (1992).
This method in our tests Campanacci (2000) always demonstrated its robustness. It

22
0.15
0.10
O
P cs ttion analysis
2 O
Tine [s/30]
U. Tarantino, D. Perugia et al.
----·without adaptative refinement
-with adl.ptative refinement
40
-,,
,
,
,

.
60
Figure 7. Error .vs.time
should point out thaL the formulation hereiu presented assumes w -1- oaud a -1- O. The
reader is referred to the original paper for the handling of such cases.
• Computatiou of the angular velocity vector {w}
1. Compute
1 n
{Po} = - "'" {p;} ,n~
i=1
1 n
{vo} = - L {ii;} ,
n
i=1
1 n
[V] = - L {Vi}{Pi - po}T
n i=l
[Xl = ~ t {Pi - pol {Pi - po}T,
n
i=1
(6.4)
(6.5)
(6.6)
(6.7)

2. Solve the system of equations
with respect to thc cartesian components wx,wy,wz of vector {w}.
3. Compute
{w}
{u} = M '
s= {u}T {va} ,
{r} = {jJa} + [w] {va}
w2
• Computation of the angular acceleration vector
1. Compute
[H] = [3]- [n2J [X]
2. Solve the system of equations
with respect to the cartesian components ax, ay,a z of vector {a}.
3. Compute (Point Q is in this case the center of acceleration).
{a}
{t} = M'
{q} = {Pa} - [a + n2] -1 {aa}
(6.8)
(6.9)
(6.10)
(6.11)
(6.12)
(6.13)
(6.14)
(6.15)
(6.16)
(6.17)
(6.18)
(6.10)

7 Conclusions
According to the resu1t8 of the numerical experiments, the method of Gupta and Chuta-
kanonta proved to be more accurate than the others.
In particular, with reference to Figures 2 and 3, the second type ofrefinement greatly
improves the accuracy of versor {u}, but deteriorates the one of the rotation ang1e ~}.
Similarly, the method of Gupta and Chutakanonta with no refinement shows the best
performance for the computation of ~), but is the worst as far as the accuracy of {u}.
The accuracy of vector {Ii} seems vcry little infiuenced by the adopted method.
The suggested adaptative approach offers an improvement of the accuracy within the
range of 1%-3%. Its main advantage is the easiness of implementation, but computing
time is doub1ed. This is almost not perceived by the user since, on the averagc, with a
PC equipped with a Pentium with 100 Mhz processor, the example presented required
about 1/100 of a second of CPU.
Bibliography
E. Pennestrl, E., A. Renzi, P. Santonocito, Dynamic modeling of the human arm 11Iith
video-based exper'imental analysis, Multibody System Dynamics,7:389-406, 2002.
G.G. Mozzi del Garbo. Discorso matematico sopra il rotamento dei corpi. Neaples, 1763.
O. Bottema, B. Roth. Theoretical Kinematics. North Holland, Amsterdam, 1979.
C.H. Suh, C.W. Radcliffe. Kinematics and Mechan'lsrns Design. John Wiley and Sons,
New York, 1978.
Pennestrl, E. Technical and Computational Dynarnics. Casa Editrice Ambrosiana, Mi-
lano, 2001. (in italian)
H. Cheng, K.C. Gupta. An Historical Note on Finite Rotations, ASME Journal of Applied
Mechanics, 56:139-145, 1989.
F.E. Veldpaus, H.J. Woltring, L.J.M.G. Dortmans. Least Squares Algorithm for thc
Equiform Transformation from Spatial Marker Coordinates, Journal of Biomechanic8,
21:45-54, 1988.
C.W. Spoor, F.E. Veldpaus. Rigid Body Motion Calculated from Spatial Coordinates of
Markers, Journal of Biomechanics, 13:391-393, 1980.
H..T. Woltring, R. Huiskes, A. de Lange, F.E. Veldpaus. Finite Control and Helical Axis
Estimation from Noisy Landmark Measurements in the Study of Human .Toint Kine-
matics, Journal of Biomechanics, 18:379-389, 1985.
T.A.G. Heeren, F.E. Veldpaus. Optical System to Measure to End Effector Position
for On-Line Control Purposes, International Journal of Robotics Research, 11:53-63,
1992.
A.J. Laub, G.R. Shiffiett. Linear Algebra Approach to the Analysis of Rigid Body Dis-
placement from Initial and Final Position Data, ASME Journal of Applied Mechanics,
49:213-216, 1982.
G.R. Shiffiett, A..T. Laub. The Analysis of Rigid Body Motion from Measured Data,
ASME Journal of Dynamic Systerns, Measurement and Control, 117:578-584, 1995.
Gupta K.C., Chutakanonta P. Accurate Determination of Object Position from Imprecise
Data, ASME J07Lrnal of Mechanical Design, 120:559-564, 1998.

Beggs, J.S. Kinematics. Hemisphere, Washington D.C., 1983.
Gupta, K.C. Measures of Positional Error for a Rigid Body, ASME Journal of Mechanical
Design119:346-348, 1997.
Park, F.C. Distance Metrics on the Rigid-Body Motions with Applications to Mechanical
Design, ASME Journal of Mechanical Design, 117:48-54, 1995
Sommer, H.J. Determination of First and Second Order Instant Screw Parameters from
Landmark Trajectories ASME Journal of Mechanical Design, 114:274-282, 1992.
Angeles, J. Spatial Kinematic Chains. Springer Verlag, Berlin, 1982.
Campanacci, G. Review and development of algorithms for the computattion of screw
parameters from experimentally measured motions, Tesi di Laurea, Universitâ degli
Studi di Roma Tor Vergata, 2000. (in italian)
Angeles J. Automatic Computation of the Screw Parameters of Rigid Body Motions. Part
1: Finitely-Separated Positions, ASME Journal of Dynamic Systems, Measurement
and Control, 108:32-38, 1986.
Angeles, J. Automatic Computation of the Screw Parameters of Rigid Body Motions.
Part II: Infinitesimally seprated Positions, ASME Journal of Dynamic Systems, Mea-
sW'ement and Control, 108:39-43, 1986.
Angeles, J. Computation of Rigid-Body Angular Acceleration from Point-Acceleration
Measurements, ASME Journal of Dynamic Systems, Measurement and Control
109:124-127,1987.

Mechanical ventilators and ventilator testers
G.Belforte, G.Eula*, T.Raparelli
POLITECNICO DI TORINO
C.so Duca degli Abruzzi, 24 -10129 TORINO
(*Tel. 011 5646911 - Fax 011 5646999 - E-mail: gabriella.eula@polito.it)
Abstract. In the human organism, the respiratory function is involuntary and essential to
life. At times, however, as in surgical operations using general anesthesia or as a result of
respiratory insufficiency, the patient needs help breathing. In the first case, the general
anesthesia completely stops the thoracic musc1es and a mechanical ventilator is needed in
order to force the oxygen-air mix into the patient's lungs (volume control ventilators). In
the second case, the patient is conscious and can breath spontaneously, but appropriate
respiratory training is useful to increase his or her pulmonary efficiency (pressure control
ventilators). The paper presents a prototype of a fully pneumatic gas-powered portable
volume control ventilator, together with two types of ventilator tester which simulate
breathing capacity and resistance of infants, children and adults. Specially developed
software makes it possible to control and monitor all respiratory parameters. AII
prototypes performed well, demonstrating the feasibility of developing new breathing
systems and testers.
Abstract. Nel corpo umana la funzione respiratoria eun atto involontario indispensabile
per la sopravvivenza dell'organismo. Tuttavia in interventi chirurgici oppure in
insufficienze respiratorie, il paziente deve essere "aiutato a respirare" con appositi
apparecchi medicali. Nel prima caso l'anestesia totale blocca completamente i muscoli
toracici e quindi la macchina deve insufflare forzatamente nei polmoni una miscela di aria
ed ossigeno (respiratori volumetrici). Nel secondo caso la persona ecosciente, e quindi
respira ancora da soIa, ma deve recuperare parte delia sua capacita polmonare con
un'adeguata "ginnastica respiratoria" (respiratori pressurimetrici). Il presente lavoro si
propone di presentare il prototipo di un respiratore volumetrico portatile di emergenza
completamente pneumatico e due modelli di tester per ventilatori. 1 tester per ventilatori
simulano capacita e resistenze polmonari di neonatilbambini/adulti. Un software dedicato
consente la verifica di ogni parametro respiratorio controllato. 1 risultati ottenuti SUl
prototipi qui presentati sono buoni e dimostrano l'efficienza di modelli innovativi.
1 Introduction
Mechanical ventilation (or artificial ventilation) began to gain ground in 1934 as a means of
providing better operating methods in anesthesiology. Initially restricted to chest surgery

28 G. Belforte, G. Eula and T. Raparelli
(where the technique was perfected), mechanical ventilation was later recognized as a more
effective way ofdealing with pneumothorax and as a means ofcompensating for the paralyzing
effect of curare on the respiratory muscles. Its effectiveness is demonstrated by the patient's
perfect oxygenation, which means that problems such as hypertension, sweating, tachycardia
and bleeding caused by respiratory insufficiency regress and disappear with great rapidity
thanks to the use of mechanical ventilators (see Mead and al. (1964), Jain and al. (1970),
Belforte and al. (1999) for arguments). These considerations apply in particular to the
mechanical ventilators used in anesthesia or in resuscitation. Such machines, in fact, chiefly
control the volume of air and oxygen delivered to the patient - hence the name volume control
ventilators. There are, however, other types known as pressure control ventilators, where the
clinician sets the operating pressure and the patient, who in this case is conscious, can use the
machine to perform respiratory exercises for therapeutic aids. Similar techniques can also be
used for athletes in training or physical therapy.
Mechanical ventilation can be classified as follows. In Controlled ventilation the work of
breathing is performed entirely by the machine. The patient cannot modify the parameters set
by the clinician. This method is typically used under general anesthesia. In Assisted ventilation
part ofthe work ofbreathing is performed by the patient, and part by the ventilator. Respiratory
parameters do not remain unchanged as in the previous mode. In High1'requency ventilation
the primary goal is to achieve gas exchange at the alveolar level through diffusion, rather than
through convection as in the earlier methods.
The most widespread method in clinical practice is controlled ventilation. With pressure
controlled ventilators, the pressure reached in conscious patient alveoli is controlled to improve
respiratory capacity. Some of interesting human physical characteristics are: vital capacity
(adult males 3.3 1; adult females 1.91); total lung capacity (adult males 61; adult females 4.21);
respiratory volume (61/min at rest); alveolar ventilation (4.2 Vmin at rest); maximum voluntary
ventilation (125-170 l/min).
To check ventilator performance for quality control and certification purposes it is
necessary to develop testers capable of simulating the respiratory system. A number of
experimental and theoretical lung models are presented in literature. The most common types
are mechanical models, which make it possible to simulate respiratory systems in normal and
pathological conditions. In general, they are practical embodiments of lumped parameter
models, where airway resistance is reproduced with calibrated holes. Lung volumes are
simulated by means of rigid or deformable containers. (see Herzog and al (1968), Lyager
(1968), Brown and al. (1964), Belforte and al. (1983) for arguments).
It should be noted that testers simulating respiratory system are current1y investigated and
constructed on the basis of international standards (see ASTM and ISO standards). A diagram
of a tester (given in ISO 10651-1) is shown in Figure 1. It consists of: a lung volume (1), a
pressure transducer (2), an airway resistance (3), a flow meter (4), and tubes of appropriate
diameter and length (5), ventilator under test (6). The ISO standard also specifies the type and
values of pulmonary or airway resistance (R), tidal volume (VI)' compliance (C) and
respiratory rate (f) to be used in prototype construction as illustrated in Table 1, as well as
providing guidelines for the length and diameter of the tubes used to connect the lung to the
ventilator under test.

Mechanical Ventilators and Ventilator Testers
1
Figure 1. Scheme of tester ventilators
2 Portable volume control ventilator
Table 1. Pulmonary resistance and
capacitance by standards
Category Adults Children
29
Infants
R (Pa s/m3) 5*10 15 20*1015 50*1015
Vt (dm3) 0,5 0,3 0,3
C (dm3/mbar) 0,050 0,020 0,001
f (cycle/min) 10 20 30
A fully pneumatic gas-powered portable emergency ventilator was developed and analyzed
(Figure 2). Portable emergency ventilators are life support devices used on ambulances,
helicopters, etc., to resuscitate accident victims. Specifically, the ventilator discussed herein
(see Belforte and al. (1992-1994) for arguments) is a time/patient cycled volume control
system which can be used on both children and adults. It is provided with three alarms (mask
loss, maximum pressure and gas cylinder consumption), while respiratory parameters can
adjusted through a wide range ofmodes and settings. Thus, it is possible to regulate respiratory
rate from 3 to 80 cycle/minute, I:E ratio (inspiration time/expiration time) from 1/5 to 2/1
steplessly. In addition, flow rate can be regulated according to calibrated scales in such a way
that the I:E ratio can be varied without affecting the tidal volume delivered to the patient.
A complete schematic view of the ventilator is shown in Figure 3. The ventilator includes
an inlet gas filtration unit W and pressure reducer B, an ON-OFF selector switch A, two
pressure reducers C and D which supply the control circuit, a control logic circuit, a patient
demand sensor J adjusted extemally via resistance 1, a selector switch K to deactivate the
patient demand sensor when desired, a supply pressure alarm sensor E, a minimum pressure
(mask loss) alarm sensor U, a maximum pressure alarm sensor T adjusted extemally, a patient
delivery valve P triggered by the control circuit, a vacuum generator Q, a 40% ~ valve X, a
100% ~ valve Y, a 40%/1 00% ~ air mix selector switch R, a pressure gauge S installed on
the outer panel to enable the clinician to monitor pressure, and a non-rebreathing valve V
connected directly to mask Z so that the patient can exhale used gas.
The ventilator can operate with a compressed air source or with oxygen. In the latter case,
enriched air can be delivered to the patient. Oxygen from supply source (3 bar) passes through
filter W and reducer B and reaches the ON-OFF selector A. When it is set to ON position, gas
is delivered to the control logic circuit, supplied at 1.4 bar and 0.8 bar by means of C and D.
The inlet pressure alarm sensor E compares 3 bar and 1.4 bar pressures. When the former drops
too far, the nozzle in valve E opens and the 1.4 bar pressure causes a metal reed to vibrate,
generating an audible alarm signal (90 dB(A) a 0.5 m).

Figure 2. Pneumatic portable volume control ventilator prototype
The output signal from the control logic circuit (uT2) provides the pneumatic command for
valve P in accordance with the respiratory rate f and the I:E ratio set by means ofresistances G
and L. In addition, the uT2 signal is used to supply the minimum pressure alarm sensor U and
to close the pressure-controlled non-rebreathing valve V.
The minimum pressure alarm sensor U is provided with a diaphragm separating the nozzle
connected to the control circuit from the chamber connected to the mask. If the patient loses
the mask and the pressure drops below 5 mbar, the jet issuing from the nozzle causes a metal
reed to produce an alarm signal (90 dB(A) at 0.5 m).
Valve V permits the gas exhaled by the patient to be discharged. The uT2 signal described
above need be sent to it only in special circumstances, e.g., when applying PEEP.
The single signal u from the control circuit is used to supply the patient demand sensor J,
which is capable of detecting if the patient recovers consciousness, generating a vacuum
greather than or equal to -2 mbar. In this case the diaphragm valve moves downwards and the
control circuit signal is discharged to the atmosphere, allowing the inspiratory phase to begin
before the set time.
The gas to be delivered to the patient crosses valve P and reaches the vacuum generator Q,
where pressure drops from the 3 bar supply level to the 20-30 mbar, useful for human lungs.
An air-oxygen mix is also generated in vacuum generator Q, where the amount of oxygen
enrichment is established by the two valves X and Y, activated by means of selector switch R.
If valve X is active, the primary ~ jet vacuum generator draws air from outside, forming a
40% O2 mix. If valve Y is active, no air can be drawn from outside, and 100% ~ is delivered
to the patient. Vacuum generator output is also connected to maximum pressure sensor T.
When the patient circuit reaches maximum pressure setting, the gas jet issues from the nozzle
and strikes a metal reed, which generates an audible alarm signal (78 dB(A) at 0.5 m). Ali
alarm sensors are designed in full compliance with current ASTM and ISO standards and, in
line with the need to develop a completely pneumatic system, are free from electric
components.
The control circuit shown schematically in Figure 3 consists of four micropneumatic
elements (pneumatically controlled monostable 3-way valves with two operating positions).
Maximum operating pressure is 1.4 bar. The first two elements F in this circuit form a
pneumatic oscillator, while the third is a threshold trigger N and the fourth element °performs

Mechanical Ventilators and Ventilator Testers 31
a logical AND operation on signals generated previously (figure 4). Output from the first
element F is connected to the next element via a delay line consisting of resistance G and
capacitance H, through which the period ofthe signals can be adjusted. Signal u is then sent to
trigger N control chamber: trigger's operating threshold can be fixed by resistances M and
extemaUy modified by the variable resistance L. Trigger output signal T2 and oscillator output
signal u are sent to element 0, which performs a logical AND operation. In this way, the signal
obtained from the entire control circuit is present only if u and 12 are both present at the same
time. This signal's duty-cycle is adjusted by means oftrigger resistance L, while its frequency
depends on resistance G and capacitance H.
Experimental tests carried out on the prototype demonstrated its serviceability, showing
that aU respiratory parameters listed above are fuUy adjustable (see Belforte and al. (1992) for
arguments).
Figure 3. Ventilator inside circuit
3 Mechanical ventilator tester
Figure 4. Control signal elaboration in
ventilator logic control circuit
When rigid reservoirs are used, models must be constructed with very large volumes (SO dm3
in the case of the adult lung model) in order to satisfy the required parameters realising
respiratory system model with a high degree of accuracy, but bulky and ill suited for use
outside laboratory.
Figure Sa shows the variable capacity tester model (see Belforte and al. (2000-2001) for
arguments). It is relatively light and compact, and can thus be used for on-site checks on
mechanical ventilators. In this prototype (see Figure Sb) the capacity variations are achieved by
means of a compliant rubber beUows (l) whose stroke is limited by three appropriately sized

springs, A-B-C, in order to simulate the lung capacity of an adult, a child and an infant. The
springs are arranged in series and separated by suitable spacers (2). Volume selection is
accomplished by means of three shafts (3), each provided with two cams (4) located on two
different levels and out ofphase by 1800 •
Appropriately controlled, the cams block either plate (5) (volume 1 = infants) or plate (6)
(volume 2 = children). When both plates (5) and (6) are free to move, volume 3 = adults is
selected. Rotational movement of (3) is provided by an electric motor. Drive is transmitted
from one shaft to the others by a toothed belt (7) and three pulleys. Volume selection, sensor
reset, a number of parameter settings and experimental tests are controlled by specially
developed software resident in the tester. The unit is controlled from a console consisting of a
backlit display and a 20-key keypad. This keypad can be used to select the various types of test
performed by the machine (Normal test; Autorepeat test; Assist test; Trend test;
Leak/Compliance test), with the airway resistance aud volume corresponding to each of the
three cases (adults, children, infants).
~ ~
Figure 5. Fixed-capacity ventilators tester prototype
In the Normal test can be controlled these parameters: ventilation frequency; IIE ratio; tidal
volume; minute volume; inspiratory time; expiratory time; expiratory plateau; cycle time; max
airway pressure; max lung pressure; PEEP; mean airway pressure; inspiratory flow; expiratory
flow. The Autorepeat test automatically repeats and prints the same parameters as the Normal
test at a predetermined time interval programmed from the keypad. The Assist test evaluates
the sensitivity of the respirator connected to the model during assisted ventilation. The Trend
test is designed to evaluate all variations in tidal volume above a certain threshold programmed
from the keypad. The Leak/Compliance test is used to evaluate any leaks in lines connecting
the tester to the ventilator or in the ventilator itself.
The resistances were constructed from commercial tubes differing in number, internal
geometry and length. Characteristics were determined using the test bench shown in Figure 6a,
where R denotes the resistance under test, S the compressed air supply, A a pressure reducer, F

a flow meter, CI and C:z the pressure measurement tubes (constructed as per ISO 6358), B and
D two H20 manometers. The characteristic curves obtained by plotting pressure drop IIp versus
flow rate Q are shown in Figure 6b. Their linearity indicates laminar flow in each resistance,
the numerical value for which coincides with the angular coefficient of the line interpolating
the test points. Error can be regarded as acceptable, as it is within the limits established by the
standards (10 %).
R
-
"' 300
Q.
-; 250
u
~ 200
OI
~ 150 +-'~-r----r---~~~--
"ti
~ 100
"'~ 50
Q.
O +-~=T----~--~--~----+---~
0,00 0,02 0,04 0,06 0,08 0,10 0,12
Q (dml/s ANR)
a) b)
Figure 6. Airways resistance: a) experimental test-bench; b) results obtained
The flow meter was calibrated using the test bench shown schematically in Figure 7a.
2000 -Trans. output slgn_1 (mV) 1------, 1,6
- Parabollc dl_gram
~ 1600 'r---=-:'::<:.J:"'F= '-==';7-
'u;
1:;;1200
- E
g -800
oi
c:
I~ 400
1,4 ~
.o
1,2 .s
1,0 ~
:>
0,8 ~
0,6 ~
0,4 g.
0,2 Ci
o~"----.-----r----+-----+ 0,0
0,0 0,5 1,0 1,5 2,0
Q(dm3/sANR)
a) b)
Figure 7. Flowmeter device: a) experimental test-bench; b) results obtained
The meter consists of a series oflamina arranged perpendicular to each other in such a way
as to divide the passage into several smaller-section passages, thus creating a pressure drop IIp
(see Belforte and al. (1996) for arguments). In Figure 7a, flow meter MF is connected: to a

compressed air source S regulated by reducer R, to an H20 manometer M, to a pressure-drop
manometer D, to a pressure transducer T for measuring the differential pressure across meter
MF, to a flow meter F for measuring flow rate Q.
The characteristic curve obtained is illustrated in Figure 7b, where pressure drop is plotted
versus flow rate. In order to process these data, a matrix of the sensor voltage readings and the
corresponding dm3/s values read by flow meter F was included directly in the prototype's
software. This reduced error from 10% to 3-6% (see Belforte and al. (2001) for arguments).
This flow meter can be also used as spirometer. The tester was compared with a fixed-capacity
lung model constructed in accordance with ISO 10651-1. This reference model consists of
three airway resistances and three fixed tanks, having volumes of 50, 20 and 1 dm3 to simulate
adult, pediatric and infant lung capacities. Using both systems it is possible to compare
measured respiratory parameters, and thus determine the portable device's reliability and
measurement repeatability. The portable tester was connected to the electronic control board
and to a pneumatic square wave generator simulating the ventilator. As required by standards,
error between theorical and experimental values was less than 10-15% (see Belforte and
al.(2000) for arguments).
4 Conclusions
Studies were carried out on mechanical ventilators and on machines used to calibrate them.
The portable volume control ventilator prototype is fully pneumatic and thus suitable for use in
any type of environment. It is powered by the same fluid employed to the patient and
consequently eliminates the additional weight and bulk of the batteries required in electrical
equipments. Testers developed for ventilator calibration both showed good performance and
reliability. The portable model in particular can also be transported to hospitals for on-site
checks on ventilators. The non-portable model provides higher accuracy and is more
appropriate for laboratory testing and for special certification purposes.
Appendix A
• I:E Ratia (inspiratarv-expiratarv ratia): inspiratory and expiratory phases ratio durations. The latter
phase is passive and usually longer. With mechanical ventilation, I:E ratios may be greater than 1.
• ridal valume (Vt): amount of air flowing in or out of the lungs during a normal or passive inhalation
or expiration (approximately 0.5 dm3 in adult males), measured with a spirometer.
• Vital capacity: sum of tidal volume, inspiratory reserve volume and expiratory reserve volume.
Generally between 3500 and 5000 mI. Depends on the patient's physical condition.
• Total capacity: Vital capacity plus residual volume. Can be estimated at around 6000 mI.
• Compliance: A measure of respiratory system's distensibility. It is defined as the relationship
between the volume and the pressure of the gas in the alveoli and is associated with lungs elastic
properties. In a fixed capacity model, compliance is (1), in a variable capacity model is (2).
C = dV (1) C = W" +a +2a. !J.p (2)
$ ~ ~
(V= air volume introduced in the model, p= model intemal pressure used, Wo = receiving container
initial volume; Pa = atmospheric pressure; a =container compliance; l1p =tank pressure change).

• Airway resistance: defined in (3). (L'1Pw = pressure drop airway resistances, Q=airways flow rate)
R = L'1pw (3)
Q
References
Mead, J., and Milic-Emili, J. (1994). Theory and methodology in respiratory mechanics with glossary of
symbols. Handbookofphysiology. American Physiology Society, Washington.
Jain, A.B., Choukroun, M.L., Tabka, Z., Ultman, J.S. (1970). High-frequency oscillatory pressure-flow
relationship in the airways of laringo-tracheo-bronchial tree casts. Medical & Biological Engineering
& Computing.
Belforte, G., Eula, G., Raparelli, T. (1999). La pneumatica per respiratori artificiali", Oleodinamica-
Pneumatica, ed.Tecniche Nuove, 46-54.
Herzog, P., Norlander, O.P. (1968). Distribution of alveolar volumes with different types of positive
pressure gas-flow patterns. Opusc.Med.Bd.
Lyager, S., (1968). Influence of flow pattern on the distribution of respiratory air during intermittent
positive-pressure ventilation. Acta Anaesthesia Scandinav.
Brown, J., and Campell, D. (1964). The electrical analogue for lung function. Biomechanics and Related
Bio-Engineering Topics. Kenedi.
Belforte, G., and Rossetto, M. (1983). Studio di un circuito di prova per l'analisi di respiratori artificiali.
Tecnica Ospedaliera.
ASTM F1161. (1988) Minimum performance and safety requirements for components and systems of
anaesthesia gas machine.
ASTM F1208. (1989). Minimum performance and safety requirements for anaesthesia breathing systems.
ASTM FI 100. (1990). Ventilators for use during critical care.
ASTM FIlO1. (1990). Ventilators for use during anaestesia.
ASTM F920. (1993). Resuscitators for use with humans.
ISO 6358. (1989). Pneumatic fluid power. Components using compressible fluids. Determination of flow-
rate characteristics.
Belforte, G., Eula, G., Raparelli, T. (1992). Pneumatic Control of a Portable Artificial Respirator.
IFToMM-jc International Symposium on THEORY OF MACHINES AND MECHANISMS"
Nagoya (Japan). VoI. 1, 413-417.
Belforte, G., Eula, G., Ferraresi, C., Sorli, M., Raparelli, T. Patent T092AOO 0385. N.01263124,
classif. A61M. (1992). Respiratore artificiale portatile volumetrico con rapporto tra inspirazione ed
espirazione variabile.
Belforte, G., Eula, G. (1994). Analisi di prove funzionali di elementi micropneumatici utilizzati come
trigger. Oleodinamica & Pneumatica, ed.Tecniche Nuove,72-80.
Belforte, G., Eula, G., Raparelli, T. (2000). A tester for artificial respirators. MEASUREMENT-
Journal ofInternational Measurement Confederation - IMEKO, 27-200, Measurement 27,241-250.
Belforte, G., Eula, G., Raparelli, T. (2001). Macchine per la misura delle caratteristiche dei
respiratori artificiali. Oleodinamica-Pneumatica, ed.Tecniche Nuove, 54-60.
Belforte, G., Eula, G., Raparelli, T. Patent TO 96AOOOOI5, N.01284315, classif. GOlF. (1996).
Misuratore di portata diun fluido a pareti sottili.

Cardiovascular and Metabolic Effort in a World Class Sailor
at Different Wind Velocities
Tanja Princi*, Carlo Capelli**, Giorgio Delbello***, and Larissa Nevierov*
*Department of Physiology and Pathology, University ofTrieste, Trieste, Italy
**Post-Graduate School ofSports Medicine, University ofUdine, Udine, Italy
***Post-Graduate School ofSports Medicine, University ofTrieste, Trieste, Italy
Abstract. The physiological demands of sailing are highly specific, varying with wind conditions,
type of craft, and role in the crew. Upwind and downwind sailing involve different types of muscles
with predominant isometric or isotonic contraction. The purpose of the present study was to
determine the cardiovascular and metabolic effort in one world class athlete (Europe Olympic Class)
during training and regattas at different wind velocities. Heart rate (HR) was recorded at rest and
during sailing, by using Polar Vantage HRmonitor. VO, was evaluated by using Cosmed K4 (in
water), and Cosmed Quark b, (in laboratory). The results indicate an increase of HR and VO,
consumption as a function of the wind velocity. HR increased in upwind sailing more than in
downwind sailing at wind velocities ranging from 2 m/s to 7 m/s, and in downwind sailing it
increased more than in upwind sailing at wind velocities ranging from 8 m/s to 12 mls. HR and VO,
consumption were larger during regattas than during training at ali the evaluated wind velocities. The
mental stress could be considered as a relevant factor during regattas, compared with training at the
same wind velocities. At high wind velocities (from 8 m/s to 12 m/s) the specific and highly
demanding craft position in downwind sailing, associated to the mental stress, could be interpreted as
a cause for the higher HR values compared with the values registered during upwind sailing at the
same wind velocities.
(Keywords: Dinghy sailing; Europe Olympic Class; Training, regatta; Wind velocity.)
Introduction
The physiological demands of sailing are highly specific, varying with wind conditions, type of
craft , and crew position (Shephard, 1990). In particular, in dinghy sailing the wind and wave
conditions (Vogiatzis et al., 1994), the time spent on the sailing course (Pudenz et al., 1981),the
frequency of the manoeuvres performed (Gallozzi et al., 1993), and the level of competition
(Gallozzi et al.,1993; Vogiatzis et al.,1994) play an important influence on the physiological
variables. Furthermore, upwind and downwind sailing involve different type of muscles with
predominant isometric or isotonic contraction. Among the manoeuvres the dinghy sailor
performs, hiking is the special posture used in order to counterbalance the capsizing effect of
the wind on the boat in upwind sailing. In hiking a sustained isometric (or quasi-isometric)
effort is performed (Felici et al.,1999), involving quadriceps and abdominal muscle groups. On
the other hand, in downwind sailing the upper limbs and trunk muscles are involved in a
prevalent dynamic activity.

38 T. Princi, C. Capelli et al.
In any case, dynamic force is less important to the sailor than a sustained isometric effort
(Shephard, 1990). In general, aerobic capacity is only moderately taxed in dinghy sailing
(Spurway and Bums,1993; Vogiatzis et al., 1994; Vogiatzis et al.,1995). In contrast, cardiac
function is challenged proportionalIy more (Spurway and Bums, 1993; Vogiatzis et al., 1995;
Felici et al., 1999). Moreover, several studies demonstrated a linear increase in
cardiorespiratory (Pudenz et al., 1981; Stieglitz, 1993) and metabolic (Vogiatzis et al., 1995)
requirements of dinghy sailing with rising wind velocity.
The purpose of the present study was to assess the cardiovascular and metabolic effort in a
world class dinghy sailor (Europe Olympic Class) during training (upwind and downwind
sailing) and regattas at different wind velocities.
Methods
The study was performed in one dinghy female world class sailor (Europe Olympic Class),
aged 27 years. Heart rate (HR) was recorded at 5 sec intervals at rest as well as during training
(upwind and downwind sailing) and regattas at different wind velocities, ranging from 2 m/s to
12 m/s, by using Polar Vantage HRmonitor. The registered data were transferred to computer
by using Polar Advantage interface and anlysed by means of Polar Precision Performance 2.0
software. HR (mean value +/- SD) was caculated on 5 min intervals of registration during
training as well as during regattas. V'02 was evaluated by using Cosmed K4 (in water) during a
training session performed with wind speeds of 3 and 7 m/s. The linear relationship between
V'02 (Cosmed Quark b2) and HR was experimentalIy determined on the subject pedalling at
50, 100 and 150 W in a separate occasion in the laboratory. This made it possible to estimate
the O2uptake during dinghy sailing from the HR values recorded during simulated regatta.
Results
During training the HR increased more in upwind sailing than in downwind sailing at wind
velocities ranging from 2 m/s to 7 m/s, and in downwind sailing it increased more than in
upwind sailing at wind velocities ranging from 8 m/s to 12 m/s. HR was larger during regattas
than during training at almost alI the evaluated wind velocities (Figure 1). Figure 2 represents
the tachograms registered at 5 sec intervals during a regatta and upwind sailing at the same
wind velocity (2 m/s).
The results indicate an increase of HR and O2 uptake as a function of the wind velocity. At 7
m/s HR and V'02 consumption reached higher values in a simulated regatta than in training
(Table 1). V'02 cyclo-ergometer values were estimated substituting the HR values recorded in
the simulated regatta in the linear regression relating HR and V'02 obtained in the laboratory.

Cardiovascular and Metabolic Effort in a World Class Sailor at Different Wind Velocities 39
170
160
'i
150
::l
140'i
> 1301:
~
120QI
.s 110
e-e. 100
~
90o::
--UPWIND
--DOWNWIND
-REGATTAS
:r
80
70
60
O 2 3 4 5 6 7 8 9 10 11 12 13
WINO VELOCITY (m/s)
Figure 1. Mean HR values registered during training (upwind and downwind sailing) and regattas at
different wind velocities (from 2 mls to 12 mls).
Wind Sailing V'02 in water V'02 cyc1o-ergometer %
(mls) conditions HRmax
ml/min % ml/min %
V'02 max V'02 max
3 Upwind 1030 34.66 % 690.9 23.25 % 54%
3 Downwind 1270 42.74 % 569 19.15 % 48%
7 Upwind 1440 48.46 % 1256.5 42.59 % 68%
7 Downwind 1440 48.46 % 1075.5 36.19 % 64%
7 Regattas 2070 69.67 % 1829.1 61.56 % 81 %
Table 1. V'02 consumption (ml/min, % V'02 max) and HR (% HR max) measured in water (V'02 in
water) (during training and a simulated regatta), and estimated (V'02 cyclo-ergometer) by using
the linear regression relating V'02 to HR obtained on cyclo- ergometer.

40
,..
'"
Rcgatta - Training
'"Tlll1c(mlll)
T. Princi, C. Capelli et al.
l'
-UII}.I'}I
"~41-.1 ,.~
Figure 2. Tachograms registered at 5 sec intervals during a regatta (20 min of registration) and training
(upwind sailing) at the same wind velocity (2 m/s). Regatta - mean HR: 116 bpm. Training -
upwind sailing - mean HR:83 bpm.
Discussion
This study represents the cardiovascular and metabolic effort in one high performance female
dinghy sailor (Europe Olympic Class) at different wind velocities during training (upwind and
downwind sailing) and regattas. The results show an increase of cardiac and metabolic
requirements of dinghy sailing with rising wind velocity in agreement with Vogiatzis et al.
(1995). In particular, during training upwind sailing is correlated to a higher increase of HR
than downwind sailing at wind velocities ranging from 2 m/s to 7 m/s (Figure 1). This is
probably due to the hiking posture and therefore to the prevalent isometric exercise in upwind
sailing. Therefore, it seems to be the degree of isometric effort, required to counterbalance a
dinghy in these wind speeds, which principally determined the cardiovascular demands, as
illustrated by Vogiatzis et al. (1995). Moreover, these data confirm the conclusions of other
Authors (Maas et al., 1989; Rowell and Shepherd, 1996), which reported at the same oxygen
uptake a higher HR in static exercise compared with dynamic exercise. On the other hand, at
stronger wind velocities (ranging from 8 m/s to 12 m/s) downwind sailing provoked higher HR
values (Figure 1). This observation suggests a strenuous dynamic activity of upper Iimbs and
trunk muscles in downwind sailing at strong wind velocities for counterbalancing the force of
the wind on the boat in these conditions. In any case, in regattas HR reached in almost all
registrations higher values in comparison to the training conditions (upwind and downwind
sailing) at the same wind velocities. The emotion and the mental stress could be involved in
this relevant cardiac demand observed during regattas, as reported by other Authors (Delbello
et al., 2001).

Cardiovascular and Metabolic Effort in a World Class Sailor at Different Wind Velocities 41
Moreover, the results of this study suggest that dinghy sailing (Europe Olympic Class) elicited
a modest V'02 consumption correlated to higher HR increments in ali the evaluated conditions
(training and simulated regatta) (Table 1).
In conclusion, this is the first experimental work quantifying the cardiovascular and
metabolic effort in Europe Olympic Class (female dinghy sailor) at different wind velocities.
The mental stress could be considered as a relevant factor, bringing about a significant increase
of HR and V'02 during regattas as compared with training at the same wind velocities. At high
wind velocities (from 8 m/s to 12 m/s) the specific highly demanding craft position in
downwind sailing, associated to a mental stress, could be considered as a cause for the higher
HR values compared with the values registered during upwind sailing at the same wind
velocities.
References
Delbello G, Bizzarini E, Bratina F, Lamberti V, Prinei T, Seeusa R, Nevierov L. L'emozione del vento.
Valutazione delia frequenza eardiaea in giovani velisti. Sport e Medicina 3,37-41,2001
Felici F, Rodio A, Madaffari A, Ercolani L, Marchetti M. The cardiovascular work of competitive dinghy
sailing. J Sports Med Phys Fitness 39: 309-314, 1999
Gallozzi C, Fanton F, De Angelis M, Dai Monte A. The energetic cost of sailing. Med Sci Res 21, 851-
853, 1993
Maas S, Kok MLJ, Westra HG, Kemper HCG.The validity ofthe use ofheart rate in estimating oxygen
eonsumption in static and in combined static / dynamic exercise. Ergonomics 32(2). 141-148, 1989
Pudenz V, Dierck TH, Rieckert H. Heart frequency as a reflection of the length of the boat race course -
an experimental study of load imposed during Laser sailing. Deutsche Zeitschriftfur Sportmedizin
32: 192-195, 1981
Rowell LB, Shepherd JT. Handbook ofPhysiology. New York: American Physiological Society, 334-335,
1996
Shephard RJ. The Biology and Medicine of Sailing. Sports Medicine 9(2): 86-99, 1990
Spurway NC, Bums R. Comparison of dynamic and static fitness - training programmes for dinghy
sailors - and some questions conceming the physiology ofhiking. Med Sci Res 21: 865-867,1993
Stieglitz o. Fatigue and serum potassium in high performanee sailors. Med Sci Res 21: 855-858, 1993
Vogiatzis r, Spurway NC, Wilson J. On-water oxygen uptake measurements during dinghy sailing. J
Sports Science 12: 153, 1994
Vogiatzis r, Spurway NC, Wilson J, Boreham C. Assessment of aerobic and anaerobic demands of dinghy
sailing at different wind velocities. J. Sports Med Phys Fitness 35(2): 103-107, 1995

A mechanical model of the biceps brachii muscle
M.Gatti1, P. Pascolo1, N.Rovere1, M.Saccavini2
1 Laboratorio di Meccanica Funzionale, University ofUdine, Udine, Italy
2 Istituto di Medicina Funzionale Riabilitativa, 'Gervasutta' Hospital, Udine, Italy
Abstract. This contribution deals with the development of mechanical model of the bi-
ceps brachii muscIe. In order to account for the finite speed of propagation of the
activation pulses, the modeI has been conceived as an assembly of contiguous discrete
elements, each one excited independently and defined accordingly to the HiII's muscular
model and the Huxley's sarcomer assumptions. The proposed model has been verified
with reference to experimental data gathered during in vivo laboratory experiments.
1 Introduction
The human muscles can be considered as mechanical actuators able to generate at their ends only
contractive forces. Their functionality depends on their physiologic properties as well as on the
level ofactivation, the fatigue state, the current clongation and velocity ofcontraction.
Due to the inherent complexity of the muscles operation, engineers often do represent them
with simplified models which abstract roughly the underlaying phenomena, missing this way some
aspects which could be critical in some applications.
In this work a model is proposed which translate the Hill's and Huxley's assumptionsabout the
muscle rheology and physiology. The aim is to set up a framework able to represent the intera::-
tions between human muscles and mechanical devices as well as a tool useful to help the physician
in his or her diagnostic tasks.
This research was focused on a simple but representative skeletal muscle, i.e. the biceps
brachii.
2 A brief introduction to the muscle rheology and physiology
Every single muscle is structured on a scalar architecture made of connective, nervous and
vascular tissues as well as of muscular fibers. The fibers are cells of lengthened shape containing
sequences of sarcomers, which are the smallest contractive units inside the muscle. On its own,
each sarcomer consists, internally, of narrow bands of filaments of myosin and, at its ends, of
narrow bands ofthin filaments ofactin.
As a response to an external electrical stimulus, chemical links are established between the
myosin and actin mulecules with a subsequent shift of the thick filaments with respect to the thin
ones, generating the muscular contraction mechanism. On the other side, while no nervous StiIIll-
lus is applied to the muscle, the muscular force, if any, is due essentiaIly to the elasticity of the
cormective tissues ofthe whole muscle.

44 M. Gatti, P. Pascolo et al.
The smallest functional unit of the muscle is caUed motor unit and is composed by one mo-
torneuron and the muscular fibers connected to it. The motorneuron Iinks the motorunit to the
central nervous system which in turn acts its control through electrical pulses called action poten-
tials. These ones last about 1+3 ms and propagate inside the muscular fibers at a finite speed,
called conduction velocity, whose value is usually in the range of 3+6 mls. Every fiber replies to
each input pulse with a complete contraction followed by a prolapse. The duration of the whole
response varies from about 7 ms (iffast fibers were excited, i.e. white fibers, able of fast contrac-
tion but weak with respect to fatigue, due to their anaerobic metabolism) to about 100 ms (if slow
.fibers were excited, i.e. red fibers, able of slow contraction but enduring to fatigue, due to their
aerobic metabolism). The nature of the motor unit response depends on the amouIt of the slow
fibers with respect to the fast ones; the response of the whole muscle is simply the summation of
the responses ofaU its motor units.
The central nervous system controls the muscles essentiaUy by means of two mechanisms:
first, a frequency modulation of the pulses, ranging from the single pulse up to trains or rapidly
emitted pulses whose effect is the staturated superimposition ofthe single pulse responses (tetanic
conditions); second, a selective recruitment of the motor units. In the latter case, the sequence of
activation depends on the size of the motorneurons, the smaUest ones before the biggest ones, due
to the different activation threshold. This mechanisms allows a soft moduhtion of the muscular
force too, because, usually, the first recruited units are plenty of slow fibers; furthermare, it per-
mits a turnover ofthe fibers to prevent the effects ofthe fatigue, when appliable.
The force output by a muscle depends on its physiological properties, its elongation, the veloc-
ity of contraction, the activation level and the fatigue state. An example of isometric curve,
measured while keeping constant the distance between the tendineous ends of the muscle, can be
seen in Fig. 1 where both the passive static force and the active static force under tetanic stimula-
tion of aU the motor units are represented. In the passive case, one can observe as the muscular
reaction, initially negligible, increases rapidly above a threshold elongation. This is due to the
inherently non linear strain-stress characteristics of the connective tissues. In the active case, the
muscle is unable to output any farce below another threshold elongation while the maximum trac-
tion is achieved near an optimal value for the length ofthe muscle. This behaviour is related to the
degree of overlap between the myosin and actin filaments, which can help or inhibit the chemicaIs
Iinks responsible of the muscle contraction. An example of the force to velocity constitutive law
can be seen in Fig. 2 where one can observe as the finite time necessary for the set up of the
chemicallinks reduces the muscular force as the contraction velocity increases.
The plots in Fig. 1 and 2 refer to not tired conditions. As time goes by, an initially maximal
and continuative contraction decays depending on the amount of fast and slow fibers as well as on
the training and the fitness ofthe muscle under study. A similar decay can be highlighted by cyclic
impulsive contraction tests.
3 A rheologic model of the human muscle
When a muscle is activated, its contractile kernel can be identified with the chemicallinks between
the myosin and actin mulecules. an their own, these mulecules compose a sequence of filaments
(connected in series) embedded in a matrix made of connective tissues (connected in paraUel to

A Mechanical Model of the Biceps Brachii Muscle 45
the filaments). Furthermore, part ofthe latter tissues, mainly the tendons, are connected in series to
the muscular fibers.
A class ofmodels which can describe this structure was introduced the first time by Hill [1] who
represented the whole muscle as an assembly of a unique contractile element and some passive
elements.With reference to the so called Maxwell configuration, a contractile element Ce is con-
nected in series to an elastic element Se, while both Ce and Se are connected in parallel to a second
elastic element, Pe, as shown in Fig 3. When the muscle is not activated, only thePe branch works.
In this paper a variant ofthe Hill model is proposed. In detail, for the parallel elementPe the fol-
lowing constitutive law was identified:
(1)
where Ppe is the action exercised by Pe, Ppo is the preload on Pe , K pe is an experimental constant
and LlLpe is the difference between the actual and the initial elongation ofpe.
In a similar way, for the series element Se the following constitutive law was used:
PSe = Pso ( e KSe ALSe - 1) (2)
The force PCe exercised by the contractile element Ce depends on the number of the activated
chemicallinks, the distance between the myosin and actin filaments and the time elapsed from the
beginning ofthe contraction. About these items, this work was inspired by the theory ofthe sliding
filaments proposed by Huxley [2]. According to this theory, once defined the fraction n of the
active links with respect to all those ones activables, the growth of the active links can be de-
scribed by means ofthe following law:
an / Ot = ( 1- n )j( u , t) - n g( u , t ) (3)
where u is the distance between the myosin mulecule (in resting condition) from its potential tar-
get, normalîzed with respect to the maximum distance at which a link is still possible, whilejand
g are fuctions that Julian [3] approximated as:
j=O
j = r( t) f] u
j=O
g=g2
g=gl U
g=g] U
if u < O
if O~ u ~ 1 (4)
if u> I
with fI, g] and g2 parameters (f] and gl depends on the preload length) while ris the so called
activationjactor. After Wong [4]-[5], rcan be expressed as a function oftime accOIding to:
(5)
where p depends on the concentration ofthe Ca++ ions anda, pand mare parameters.
In this work, the short term constants fI, g] and g2 appearing in eq. 4 and the p, a, pand m fac-
tors inside eq. 5 have been estimated according to the available literature, adding the feature of
being time dependent due to fatigue, as stated by specific experimental tests run by the authors.

46 M. Gatti, P. Pascolo et al.
Once n and u have been calculated, the istantaneous forcePce is given by the integral:
Pce=K~nudu (6)
where K is specific ofthe muscle under study andQ is the space ofthe available values foru.
FinalIy, the total muscular farce, PM = Ppe + PSe, folIows from the the equilibrium and congru-
ence conditions an the elements P., Se and Ce.
4 Numeric models for the muscIe
The equations introduced in the previous section have been tranlated into numeric models at the
Laboratorio ofMeccanica Funzionale ofthe University of Udine.
InitialIy, the isometric contraction of a single sarcomer was studied. In Fig. 4 one can see the
time varing forces as estimated by the model when considering a range of preload values. In this
application, due ta the uncomplete data available, the sarcomer length was estimated by reference
ta the whole muscular fiber length, assuming a similar behaviour between fiber and sarcomer. This
applies ta both the length at which no force can be output (Lthres,inf) and the maximal length, at
which the overlap between the myosin and actin can no longer be guaranteed (Lthres,sup)' As ex-
pected, the model generated null forces at length below Lthres,inf while at length above Lthres,sup the
behaviour was govemed by the parallel elementPe only.
In a second phase, the isometric contraction of a whole fiber was studied. The fiber was mod-
elled as an assembly of contiguous sarcomers, connected in series and each one made active
independently ([6]). The latter property was exploited in order to simulate the effects of the finite
velocity of propagation of the the action potentials along the nelVous tissues of the muscle. By
exciting in sequence the sarcomers, the model highlighted how the response ofthe single sarcomer
is not isometric; in short, as the electrical pulse propagated, the overall irometric condition was
satisfied by the balance baween the shortening ofthe already excited elements and the lengthening
ofthe not yet actived sarcomers. In Fig. 5 some snapshots are given.
FinalIy, the models herein proposed were compared, satisfactorily, ta other numeric represenn-
tions generated by means of Virtual Muscle, a third party application [7] based an assumptions
similar to those ones presented in this paper,
5 Numerical simulations on the biceps brachii
The simulations run in this study focused an the biceps brachii because of its simplicity and high
representativity among the skeletal muscles. Furthermare, the model was parametrized with refff-
ence ta an healty and trained subject.
The isometric contraction at optimal length and different levels of activation as well as the iro-
metric contraction at maximal activation and different lengths were studied.
The results obtained in the two cases, normalized with respect ta the maximal force, are repe-
sented respectively in Fig. 6 and 7. From these figures one can also notice the absence of any
fatigue effect, due ta the particular muscular parameters and the short time ofsimulation.

[Paolo b. pascolo_(eds.)]_biomechanics_and_sports

[Paolo b. pascolo_(eds.)]_biomechanics_and_sports

Recommended

Recommended

More Related Content

Similar to [Paolo b. pascolo_(eds.)]_biomechanics_and_sports

Similar to [Paolo b. pascolo_(eds.)]_biomechanics_and_sports (20)

Recently uploaded

Recently uploaded (20)

[Paolo b. pascolo_(eds.)]_biomechanics_and_sports