F. R. Rozema ~, E.

Otten 2,
Computer-aided optimization of R. R. M. Bos 1, G. Boering ~,
J. D. van Willigen 2
~Department of Oral and Maxillofacial
choice and positioning of bone Surgery, University Hospital Groningen,
Groningen, The Netherlands, and
2Department of Neurobiology and Oral

plates and screws used for Physiology, University of Groningen,

Groningen, The Netherlands

internal fixation of mandibular

fractures
E R. Rozema, E. Otten, R. R. M. Bos, G. Boering, J. D. van Willigen: Computer-
aided optimization of choice and positioning of bone plates and screws used for
internal fixation of mandibular fractures. Int. J. Oral Maxillofac. Surg. 1992; 21."
373-377. © Munksgaard 1992

Abstract. The present study describes a biomechanical integrated model of the

mandibular system in which the maxilla and mandible, the masticatory muscles,
and the temporomandibular joints are regarded as one system. In this model,
strains in plate-osteosynthesis devices for internal fixation of mandibular fractures
can be minimized by optimizing their positions. The model evaluates maximal
bite force strategies on all possible dental elements; it uses a linear programming
technique and a muscle architecture model, resulting in muscle recruitment pat-
terns. The shape of a "standard" lower jaw is digitized by means of a three-
dimensional (3-D) coordinate retrieval device and drawn on a computer screen
after its dimensions have been changed according to the clinical case at hand.
Key words: mandibular fractures; computer;
The 3-D location of the fracture as well as the anatomic restrictions for screw bone plates; osteesynthesis.
placement can be indicated on the screen. Osteosynthesis devices can be indicated
in terms of dimensions, number, and material properties. Accepted for publication 25 August 1992

For many years, O M F surgeons have
used several types of fracture fixation
devices. The main concern has always
been the strength of these plates and
screws in order to enable undisturbed
fracture healing. Studies concerning
mechanical properties and stability have
clearly demonstrated that the presently
used plate-osteosyntheses are more
stable than wire fixations 6. The AO and
Champy bone plates were found to be
more resistant to vertically bending
forces of the mandible than was wire
fixationlO,14,29.
Mathematical models, predom-
inantly referring to the mandible, have
been developed to formulate biomech-
anical justification of the positioning of
different plates and screws. On the basis
of simple mathematical models and
photoelastic analysis that uses polarized
light on araldite bars representing the
mandible, CHAMPY et al. 7 concluded
that tension zones occurred at the upper
border, and compression areas at the
lower border of the mandible. The inci-
sor and canine regions revealed a zone
of torsional moments. These obser-
vations were confirmed in other
studies 21'3°. The indications for the use
of different types of plates must, how-
ever, be assessed critically in each frac-
ture. Specific indications for mono- or
bicortical osteosynthesis have been de-
scribed8'21. For instance, comminuted
and infected fractures contraindicate
miniplates, as do fractures in jaws with
developing teeth 8.
Several studies have been performed
on the intact human masticatory system
in which the maxillae and mandibles,
the masticatory muscles, and the tem-
poromandibular joints were regarded as
one system 2J2'15'~9'22'28'?1-33. In these
studies, the optimal directions and mag-
nitudes of bite forces acting on the man-
dible in different positions in relation to
the occlusal plane were analyzed. At the
same time, some understanding about
muscle recruitment as a way to optimize
efficiency or maximize bite forces has
been provided. The generated models
have been useful because electromyo-
graphic (EMG) measurements which
can predict the possible influence of the
acting muscles on the magnitude of bite
force were hard to obtain.
In previous studies, resorbable poly-
L-lactide (PLLA) plates and screws used
in osteosynthesis have been discused 3-5.
However, the polymeric plates used re-
vealed breakage at 50% of the tensile
force which could be applied to the
Champy miniplates 3,16. As it became
clear that the mechanical properties of
PLLA plates and screws were not com-
parable to those of metal plates, the
need was felt to understand the height
and direction of the forces acting on
the human mandibular system. For that
reason an integrated computer model
was developed that could predict these
forces. This model should calculate
movements based on muscle recruit-
ment and contain the morphometrics
374 R o z e m a et al.

o f bony elements, jaw joints, and
muscles 23'24 o f the jaw system.
The aims o f this model are, firstly, to
analyze the possibilities for application
of resorbable osteosyntheses; secondly,
to understand the factors determining
the loads across a fracture and the mag-
nitude and direction o f forces on osteo-
syntheses; and, thirdly, to be able to
make a decision within a reasonable time
about the optimal localization and direc-
tion o f osteosyntheses in individual
cases.
Material and methods
The software package called "JawKit" was
written by OftEN et alY for the Apple Macin-
tosh II series of computers. The Macintosh
Programmers Workshop (Apple Computer,
Inc. 1985-8) 3.0 environment was used for
programming and debugging. Most of the
graphics was taken from a computer recon-
struction package called MacReco (Fig. 1)25.
The data on human muscle architecture were
taken from WHJS & HILLEN34, summarized by
KOOLSTRAet al. 15.
In order to digitize the three-dimensional
(3-D) surface of a "standard" human man-
dible, a digital position-recording system was
developed (Fig. 2), consisting of a rotating
table on which the mandible can be mounted,
fitted with an optical potentiometer with a res-
olution of 0.1 degrees, A pen is fixed to two
sliding bars with steel wires, and it is also con-
nected to optical potentiometers with a resol-
ution of 0.1 mm. With the pen any point on
the mandible can be indicated from which a
unique set of three cylindrical coordinates
(two positions and one angle) is shown on a
digital display. By means of an interface the
data are collected. The data can be made ac-
cessible to the graphics software.
For an optimization model as offered in the
present paper, knowledge concerning muscle
recruitment is of vital importance. As in a
number of previous studies 15'22, the mandible
has been modeled as a static system, supported
by the jaw joints and a bite point location, and
balanced by a redundant set of muscle forces.
In order to find a realistic muscle recruitment
pattern, we need an optimization criterion.
From EMG measurements we know that re-
cruitment involves a gradual increase in
muscle activity at increasing bite force in a
given bite force direction while the pattern of
muscle activity as such, does not change9.
KOOLSTRAet al. 15 based their model on these
findings and looked only for the optimal re-
cruitment pattern, using a linear program-
ming technique by minimizing the largest re-
cruitment of any of the muscles. This ensures
that a recruitment pattern is found that not
only balances the mandible, but also provides
the highest bite force attainable. Their method
was used in the present software (indicated by
"Optimal Bite Software" in Fig. 1). The linear
programming software was taken from PRESS
et al. 27.As a starting point, we chose at random
a "normal looking", Caucasian young, male
mandible with complete dentition, which was
obtained at dissection. The mandible was pre-
pared by standard maceration procedures.
The software allows digitized deformation
of the "standard" mandible, to accommodate
specific dimensions of a given case. There are
four parameters that can be changed: the
width, height, and depth of the mandible and
an angular offset in the parasagittal plane.
These parameters can be determined on trac-
ings of standardized roentgenograms by the
measurement of mandibular intergonial
breadth (gonion-gonion), mandibular height
(distance between the plane of the lower bor-
der and the tangent through the condyle paral-
lel to the lower border), mandibular corpus
length (gonion-menton), and jaw angle (be-
tween the inferior and posterior margins of the
mandible).
A user (mouse-driven) interface was written
with which it is possible to indicate the loca-
tion and shape of a random fracture. A 3-D
reconstruction of the mandible is drawn on a
computer screen from any desired viewpoint
and with any desired view angle. The user can
indicate the number, dimensions, and mech-
anical properties of the bone plates, so that
any type of bone plate can be simulated and
tested. Acceptable anchorage sites can be indi-
cated for the bone screws for fixation of the
plate.
A bone plate is simplified to a rectangular
beam element, a well-studied concept in struc-
tural mechanics. Formulas which have been
extracted from PoPOV26describe the behavior
of a rectangular beam element under complex

General System
JawMorphometrics
Software
Graphics and Interface
Physiology Force Equilibrium
across Fracture

Mo ,.ome,,cs
(jaw dimensions) 1 -
Optimizatioofn
Plate Positions

~ OptimalBite
Software ~P" I Musclpateerrecrui
ns tment~ ~mD' P"
Linear Algebra
@ ~ Fracture Location, Shape~.~.
and Orientation / - Elastic Elements
Model of Fracture
(~~ Number,Dimensionsand) Mechanicsof Elastic
MechanicalPropertiesof ~,.,,,..=,IlD~- Beams (Bone Plates)
BonePlates )

Fig. 1. Scheme of input, output, and units of "JawKit", a software package designed to optimize position and choice of bone plates in
mandibular fractures. Elliptic shapes indicate input source; rounded rectangles, input; rectangles, software; and diamond shapes, output.

3-D transducer
Fig. 2. Digital position-recording system, designed to trace the 3-D shape of surfaces of the "standard" mandible. Mandible is mounted on
platform which can be rotated (the angle (~0) of rotation is the third coordinate). A pen (P) can be moved up and down and also forward
and backward in one plane (providing the first and second coordinates, X and Y), so that any point on the mandible can be indicated.
Coordinate groups are transferred to a microcomputer.
loads as a combination of torsional, bending,
S-shaped, and linear deformations.
A major task in completing the software
w a s to design an algorithm that could calcu-
late the displacements and rotations across the
fracture at a given position of one or more
bone plates under the load of an external force
and torque. It was decided to use an elastic-
elements method designed to suit the given de-
tail of the elastic combination of bones and
plates.
First the fracture surface was divided into a
number of triangles. Each of the bony tri-
angles is considered to provide an elastic re-
sistance for its counterpart at the other side of
the fracture at right angles to its surface. This
elasticity is taken to be 19.4 GPa (measure-
ments on fresh human mandibles by ASH-
MAN & BUSHKIRK1).In this way a number of
push springs was defined in space with differ-
ent directions. Together with the bone plates
in position and in combination with their
mechanical properties, this system has elastic
energy as a function of 3-D displacements and
rotations (six degrees of freedom) at a given
external force and torque acting on the system,
A minimum in this energy function is a solu-
tion in which all forces and torques cancel each
other. This minimum was found by using a po-
lytope search algorithm 2°.
After finding the nearest minimum, the
compressing forces on the bone, the displace-
ments at the fracture, the deformations of the
bone plates, and the forces and torques gener-
ated by the bone plates were known. The
maximal internal strain anywhere in any of the
bone plates was also known; this strain is a
variable that needs to be minimized in order
to obtain a minimal chance of bone plate frac-
ture. This bone plate strain can be minimized
by looking for the optimal placement of the
bone plates, since that placement influences
the bone plate strain. Because of the rather
complex mathematical relation between maxi-
mal bone plate strain and the coordinates of
Computer and bone plates 375
B'gg~lX
interface microcomputer
placement of the anchorage points of the bone Figs. 3a and b is the result of a number
plates, a grid-search strategy was used, pre- of mathematical and graphic operations.
ceding a polytope search algorithm which cal- These operations each cause small arti-
culates the maximal bone plate strain at each
facts in the resulting image, so that small
of the grid points. After that, the point at
which the lowest maximal bone plate strain deviations in shape from the actual man-
occurs was selected as a starting point for the dible occur. However, the sizes of these
polytope search method, which looks for the deviations are within the limits of the
closest actual minimum in maximal bone plate other data (on muscles for instance) and
strain. operations (search methods) used.
For example, the mechanical prop-
erties and dimensions used were those of
Results a high-molecular-weight, as-polymer-
ized P L L A which had been tested in sev-
Fig. 3 shows the typical result of a com-
eral studies 4'5'16,17. In the case of the ser-
puterized search for optimal position o f
rated fracture, the m a x i m u m strain oc-
bone plates and screws. Muscle forces,
curring in any part of the bone plate was
joint vectors, one bite vector (of the 13
1.08%, whereas in the case o f the smooth
divided over the dental elements), and
fracture the maximal strain amounted to
loads on plates and bone at the site of
1.56%, which is beyond the tolerance of
the fracture are presented. Only one bite
the material for the plate dimensions
vector is drawn, but the optimization
used.
was done with the whole set of possible
After some trials with our software, it
bite point locations on all dental ele-
appeared that not only the absolute
ments in directions perpendicular to the
strength of bone plates but also their
occlusal plane. Robustness analyses
shape is relevant to their success: broad
showed that directions of the bite vector
f l a t plates are able very well to withstand
within a cone with an opening angle of
deformation in a plane tangential to the
30 ° and an axial direction perpendicular
surface of the mandible.
to the occlusal plane gave very similar re-
The software can also calculate frac-
sults.
ture mobility. For instance, in the ex-
Fig. 3a shows a serrated fracture, the
ample shown in Fig. 3a, an axial move-
fixation demands of which differ from
ment of 0.02 m m occurs, whereas move-
those of the smooth fracture shown in
ments in the plane of the fracture did not
Fig. 3b. The serration is able to take up
exceed 0.1 mm. This ability may be of
certain components of the torque and
help in studying the relation between
also, to a certain extent, shear forces,
fracture mobility and fracture healing.
making the fixation task of the bone
plate easier: As can be seen in the figure,
Discussion
the optimal solution for the position of
the plate is different for the two cases. We have to discuss the limitations pro-
The image of the mandible as given in duced by the assumptions and abstrac-
376 R o z e m a et al.

"J -~ Ta tions used in the software. One of these

abstractions is that, although muscles
are described as separate functional units
(the masseter and temporalis muscles are
split up into units), they still operate
along single working lines and pull at dis-
crete points of insertion. It can be shown
that this is only a problem if the fracture
runs through an area of the insertion of
a muscle.
The calculation of the mechanical
equilibrium of fracture and plates is per-
formed by an elastic-elements method
with detailed representation of 3-D frac-
ture shape, with linear elastic bone prop-
erties and minimum energy search. This
represents a fairly realistic piece of
modeling, but rests on the assumption
that bone properties are homogeneous,
as we know they are not. The properties
of cortical bone, however, have been
used. Cortical bone takes up most of the
task of compressive force resistance,
d T since a torque is usually also present at
", ~ Ta the fracture, resulting in slight tilting of
one of the bone fragments relative to the
other, so that the spongeous bone is not
loaded. It must be stressed, however, that
the calculations are based on optimal
muscle recruitment patterns, generating
maximal bite forces 13. GERLACH et a l . 11
have already indicated that fracture pa-
tients treated with miniplates produce
only 17% of the maximal bite force 2 d
after operation, increasing to about 50%
after 4 weeks. This suggests strong inhi-
bition of the normal muscle recruitment
levels.
A last assumption to be discussed is
that the 3-D shape, orientation, and po-
sition of a fracture can be indicated
very accurately. Clearly, this depends
much on the quality of the standard-
ized, cephalometric roentgenograms
used, and on interpretations and obser-
vations during exploration made by the
surgeon. Perturbation analyses indi-
b T cated that the shape of the fracture is
of particular importance. Usually, how-
Fig. 3a. 3-dimensionalreconstruction of a human mandible with muscle forces (white arrows),
ever, the position and orientation of the
joint loads (J), and bite force (B) (black arrows). Bite forces have been scaled down by a
fracture have a tolerance of about 5
factor of 3.75 relative to the muscle forces. Characters in the reconstruction represent the
followingmuscleforces: (Md) and (Ms), deep and superficialmasseter; (Pm), medial pterygoid; mm and l0 °, respectively (generating
(Pli) and (Pls), inferior and superficial lateral pterygoid; (Ta), (Tp), and (Td), anterior, negligible differences in optimal plate
posterior, and deep temporalis. A serrated fracture (F) (dashed lingually) is indicated or/ positions and strains), but they can be
which a force and torque act as a result of the force equilibrium of muscle forces and loads. less robust in some cases.
The torque (T) is drawn as a double-headed arrow and is a result of the worst-case bite point With the software package JawKit,
selection. Only one bite point has been drawn for clarity. A bone plate is drawn by the more than one fracture can be analyzed.
computer. The optimal position of the bone plate was calculated by the software presented This can be done sequentially for each
in this paper. fracture, since, if one fracture is stabil-
Fig. 3b. Bone plate position optimized for smooth fracture. Note difference in position and
ized by a bone plate, the resulting com-
orientation, as compared with Fig. 3a.
bination of bony elements and plates is
mechanically supposed to be equivalent
to an unfractured bone. This implies

that a second fracture can be analyzed
as t h o u g h the first one were n o t there.
Experiments with the software indi-
cated that in all cases a complex load
across the fracture occurs. This results
in a failure o f predictions o f optimal
b o n e plate position based on the m a i n
load (ignoring the other load com-
ponents), such as those used by CHAMPY
et al. 7'18. A c o m b i n a t i o n o f loads
(torques and forces at o d d angles to the
fracture) tends to lead to unpredictable
b o n e plate positions, because such c o m -
plex loads are h a r d to envisage in terms
o f mechanics and b o n e plate positions.
As described before, decisions on the
choice and location o f b o n e plate sys-
tems have, up to the present, been based
on models in which m a x i m u m generated
bite forces were translated into dissected
m a n d i b l e halves or bars representing the
mandible. In the present study, a m o d e l
has been developed in which it is poss-
ible to integrate the functional p e r f o r m -
ance o f the masticatory system, the indi-
vidual m o r p h o l o g y , and the presenting
fracture o f the case at hand. This results
in the capture o f the directions a n d mag-
nitudes o f the muscle forces and j o i n t
loads that are acting on the b o n e and
the b o n e plate. In advance, but prefer-
ably during the operation, the surgeon
can decide on the ideal plate, optimal
positioning, and n u m b e r o f b o n e plates.
Additionally, the d a t a o b t a i n e d f r o m
the model can be used in designing b o n e
plate shape, dimensions, a n d new ma-
terials.
Acknowledgments. We would like to thank
Prof. Dr J. M. H. Dibbets for suggesting the
design of the digital position-recording sys-
tem, and the electronics section of the Depart-
ment of Neurobiology and Oral Physiology,
University of Groningen, particularly H. Mel-
chior, for designing and building the interface.
Fig. 2 was drawn by Mr F. Klok, for which
assistance we thank him.
References
1. ASHMAN RB, VAN BUSHKIRK WC. The
elastic properties of the human mandible.
Adv Dent Res 1987: 1: 64-8.
2. BARBENELJC. The biomechanics of the
temporomandibular joint: a theoretical
study. J Biomech 1972: 5: 251~.
3. Bos RRM, BOERING G, ROZEMA FR,
LEENSLAGJW. Resorbable poly(L-lactide)
plates and screws for the fixation ofzygo-
matic fractures. J Oral Maxillofac Surg
1987: 45: 751-3.
4. Bos RRM, ROZEMAFR, BOERINGG, NI-
JENHUIS A J, PENNINGS A J, VERWEIJAB.
Bone plate and screws of bio-absorbable
poly(L-actide). An animal pilot study. Br
J Oral Maxillofac Surg 1989: 27: 467-76.
5. BOS RRM, ROZEMAFR, BOEVdNGG, NI-
JENHUIS A J, PENNINGS A J, VERW~IJ AB.
Bio-absorbable plates and screws for in-
ternal fixation of mandibular fractures.
A study in 6 dogs. Int J Oral Maxillofac
Surg 1989: 18: 365-9.
6. BRONSR. Stabiele interne fixatie bij cor-
pus mandibulafrakturen. Thesis. Gronin-
gen, University of Groningen, 1970:
84 97.
7. CHAMPYM, LODDEJP, JAEGERJH, WILK
A. Ost~osynth~ses mandibulaires selon la
technique de Michelet. I. Bases biom&
caniques. Rev Stomatol 1976: 77:569 76.
8. CHAMPY M, PAVE H-D, GERLACH KL,
LODDE JP. Mandibular fractures. In:
KR/)GER E, SCHILLI W, WORTHINGTON P,
eds.: Oral and maxillofacial traumatology
2. Chicago: Quintessence Publishing,
1986: 1943.
9. VAN EIJDEN TMGJ. Pers. comm. Aca-
demic Center for Dentistry, Amsterdam
(ACTA), The Netherlands, 1990
10. EWERS R, NIEDERDELLMANN H. Span-
nungsoptische Untersuchungen bei der
Verschraubung yon Osteosyntheseplat-
ten. Dtsch Zahnarztl Z 1977: 32:349 52.
11. GERLACH KL, PAPE H-D, NUSSBAUMP.
Untersuchungen zur Belastbarkeit der
Miniplattenosteosynthese yon Unterkie-
ferfrakturen. Dtsch Z Mund Kiefer Ge-
sichtschir 1984: 8: 363-6.
12. GREAVES WS. The jaw lever system in
ungulates: a new model. J Zool 1978: 184:
271 85.
13. HILL AV. The heat of shortening and the
dynamic constants of muscle. Proc R Soc
Lond Ser B 1938: 126: 136-95.
14. KARASZI, KGRGNDI L, SZAB6 G. Photo-
elastic stress analysis on mandibular os-
teosynthesis. Int J Oral Maxillofac Surg
1986: 15: 259-62.
15. KOOLSTRAJH, VAN EIJDEN TMGJ, WEHS
WA, NAEIJE M. A three-dimensional
mathematical model of the human masti-
catory system predicting maximum poss-
ible bite forces. J Biomech 1988: 21:
563 76.
16. LEENSLAG JW, PENNINGS AJ, Bos RRM,
ROZEMAFR, BOERINGG. Resorbable ma-
terials of poly(L-lactide). VI. Plates and
screws for internal fracture fixation. Bio-
materials 1987: 8:70 3.
17. LEENSLAG JW, PENNINGS A J, BOS RRM,
ROZEMAFR, BOERINGG. Resorbable ma-
terials of poly(L-lactide). In vivo and in
vitro degradation. Biomaterials 1987: 8:
311-14.
18. LODDE JP, CHAMPYM. Justification bio-
mGcanique d'un nouveau matGriel d'os-
tGosynth~se en chirurgie faciale. Ann Chir
Plast Esthet 1976: 21:115-21.
19. MANSOUR MM, REYNIK RJ. In vivo oc-
clusal forces and moment. I. Forces meas-
ured in terminal hinge position and as-
sociated moments. J Dent Res 1975: 54:
114-20.
20. NELDER JA, MEAD R. A simplex method
Computer and bone plates 377
for function minimization. Comp J 1965:
7: 308-13.
21. NIEDERDELLMANN H. Fundamentals of
healing of fractures of the facial skull. In:
KR/2GER E, SCHILLI W, eds.: Oral and
maxillofacial traumatology 1. Chicago:
Quintessence Publishing, 1982:125 33.
22. OSBORNJW, BARAGARFA. Predicted pat-
tern of human muscle activity during
clenching derived from a computer as-
sisted model: symmetric vertical bite
forces. J Biomech 1985: 18: 599-612.
23. OTTEN E. A myocybernetic model of the
jaw system of the rat. J Neurosci Methods
1987: 21: 287-302.
24. OTTEN E. Concepts and models of func-
tional architecture in skeletal muscle. In:
Kent BR, ed.: Exercise and sport science
reviews. New York: Macmillan, 1988:
89-139.
25. OTTEN E, VAN LEEtr~VENJ. MacReco, a
3-D reconstruction package from serial
sections and images. Eur J Cell Biol 1988:
48: 73-7.
26. Poeov ER Mechanics of materials. Eng-
lewood Cliffs, NJ: Prentice-Hall, 1978.
27. PRESS WH, FLANNERY BR TEUKOLSKY
SA, VETTERLINGWT. Numerical recipes.
Cambridge: Cambridge University Press,
1987.
28. PRUIM G J, DE JONGH H J, TEN BOSCHJJ.
Forces acting on the mandible during bi-
lateral static bite at different bite force
levels. J Biomech 1980: 13: 755-63.
29. RAHN B, CORDEYJ, PREIN J, RUSSENBER-
GERM. Zur Biomechanik der Osteosyn-
these an der Mandibula. In: Spiessl B,
Schuchardt K, eds.: Fortschritte der Kie-
fer und Gesichts-chirurgie. Bd. 19.
Stuttgart: Thieme, 1975: 37~42.
30. RALPHJP, CAPUTOAA. Analysis of stress
patterns in the human mandible. J Dent
Res 1975: 54: 814-21.
31. THROCKMORTON GS, THROCKMORTON
LS. Quantitative calculation of temporo-
mandibular joint forces. I. The import-
ance of the magnitude of the muscle
forces. J Biomech 1985: 18: 445-52.
32. THROCKMORTONGS. Quantitative calcu-
lations of temporomandibular joint reac-
tion forces. II. The importance of the di-
rection to the jaw muscle forces. J Bio-
mech 1985: 18:453 61.
33. WEIJSWA. Mechanical loading of the hu-
man jaw joint during unilateral biting.
Acta Morphol Neerl Scand 1981: 19:
261-2.
34. WEIJS WA, HILLEN B. Relationships be-
tween the masticatory muscle cross-sec-
tion and skull shape. J Dent Res 1984:
63:1154-7.
Address:
Dr Fred R. Rozema
Department of Oral and Maxillofacial
Surgery
University Hospital Groningen
Oostersingel 59
9713 EZ Groningen
The Netherlands