The 13th Conference on Mechanical Engineering CME2022

March 15-16, 2022, Bordj El-Bahri, Algiers, Algeria

Development of a Shoulder Musculoskeletal Model with Scapulothoracic Joint

K.Abderrazak a, B. Yacine a

Laboratoire de Conception des Systèmes Mécanique

EMP, BP 17 Bordj El Bahri, 16111 Alger, Algérie.
{ Abderrazak Kedadria, Yacine Benabid}
kedadria74@gmail.com

ABSTRACT

Biomechanical models of the musculoskeletal system have been developed to estimate internal loading
applied to morphological structures and neurological control, which cannot directly be measured in vivo,
from external measurements like kinematics and externals forces. Various musculoskeletal models
developed are used to evaluate and analyze upper limb movements. However, the complexity of shoulder
mechanics in combination with the movement of the skin relative to the shoulder bone makes it difficult
to measure shoulder kinematics. In this study, we have developed a model of the shoulder that includes 09
degrees of freedom representing the sternoclavicular joint, scapulothoracic articulation, and glenohumeral
joint using NMSBuilder which is specifically designed to support the rapid development of computer-
aided medical applications, integrated with dedicated simulation (OpenSim). Shoulder movements for
three tasks are analyzed by the inverse kinematics simulation using this model, we compared the model to
bone-pin and found modeled kinematics to be accurate to within 2 mm root-mean-squared error for
individual bone-pin markers across all markers and movement tasks. In addition, we evaluated the Euler
angles for the scapula joint system from the model using ISB recommendation and compared them to
Euler angles computed without the model. Our model can be used for inverse and dynamics analyses and
to compute joint reaction loads.

Keywords: Biomechanics, Musculoskeletal Modeling, Shoulder, Scapulothoracic Joint, Inverse

Kinematics Simulation.

1. INTRODUCTION

Validated computational models of the musculoskeletal (MSK) system provide a framework for
integrating anatomical and physiological data, enabling a wide range of studies in neuromuscular control.
In the last few decades, a variety of complex three-dimensional (3D) models of the entire human
musculoskeletal system [2,8,9] have been developed. However, of all these models, few describe the
upper limb. An important reason for this is the upper limb's complex kinematic structure, especially the
scapular kinematics that are difficult to characterize [3] due to the scapula’s movement below muscle, fat,
and skin. Such models ought to be complex sufficient to practically imitating the behavior of the human
musculoskeletal system. Few complex upper extremity models have been developed such as the Delft
Shoulder Model (DSM) [11], the shoulder part of the AnyBody Modeling System [2], the Newcastle
shoulder model [1], and the Swedish model [5]. De Baets et al. 2013 [3] appeared that kinematics might
be dependably measured from surface markers in stroke patients for movements with a large range of
motion and a subset of scapula degrees-of-freedom. Seth et al. 2016 [13] developed the scapulothoracic
joint model that is based on an internal coordinate joint formulation [14] that captures the
biomechanically permissible kinematics and enforces the motion of the scapula without kinematic surface
constraints. We have developed a biomechanical model of the shoulder with scapulothoracic joint using
NMSBuilder software [6] and the plugin of the scapulothoracic joint in OpenSim software [15] developed
by Seth et al. [13]. The goals of this study are to 1) create a shoulder musculoskeletal model from 3D
geometries with the scapulothoracic joint; 2) evaluate the accuracy of shoulder kinematics from the model
against bone-pin data measured during shoulder abduction, flexion, and internal rotation with 90 degrees
of abduction tasks; 3) evaluate Euler angles for scapula joint system from the model using ISB
recommendation and compared to Euler angles computed without the model. NMSBuilder is an open-
source platform for musculoskeletal modeling and simulation, the operations implemented can be used in
a workflow (Figure 1) to create OpenSim musculoskeletal models from 3D surfaces. The first step
includes an information handling to make supporting objects essentials to create models, e.g. surfaces,
anatomical landmarks, reference systems; and the second step includes the creation of OpenSim objects,
e.g. bodies, joints, muscles, and the corresponding model [6].

Figure 1: Workflow for creating subject-specific OpenSim models from 3D geometries. A) Data
processing to create supporting objects necessary for the musculoskeletal model. B) Creation of OpenSim
objects and corresponding model. [1]
The International Society of Biomechanics (ISB) recommended a standard for describing the motion of
the upper extremity [17] has been adopted by several models [2,8,9]. This standard defines the kinematics
of the upper limb, such as the scapula defined in terms of a body-fixed Euler rotation sequence for a
frame fixed to the acromion (Anglus Acromialis) of the scapula relative to a frame fixed to the clavicle,
assuming a ball joint between these segments (Figure 2-B). Seth et al. [13] characterized the
scapulothoracic joint by the scapula's translation and rotation on the thorax's surface modeled as an
ellipsoid [14]. The scapulothoracic joint defines the kinematics of a joint frame on the scapula with
respect to a joint frame on the thorax body. The scapulothoracic motion was parameterized (Figure 2-A)
by four-position coordinates: 1) abduction-adduction, elevation-depression, upward rotation, and internal
rotation or “winging”.

(A) (B)
Figure 2: (A) The four degrees of freedom of the scapulothoracic joint [13], (B) Scapula coordinate
system and definition of AC motions. Yc is the local axis for the clavicle coordinate system (Please note,
the origin, shown here at AC, should be placed at AA). [17]

2. MATERIALS AND METHODS

2.1. Geometric Model

To create our shoulder model, we used the dataset BodyParts3D (The Database Center for Life Science,
Japan). This dataset consists of a collection of surface meshes representing muscles, bones, and other
structures. Bones were performed in MeshLab using smoothing and decimation processes to improve the
poor quality of the mesh.
2.2. Shoulder Musculoskeletal Modeling using NMSBuilder Software
We applied the described workflow for subject-specific OpenSim models (Figure 1) to create a multibody
model of the shoulder that included an articulated linkage with:
- 4 rigid bodies: thorax, clavicle, scapula, and humerus.
- 9 degrees of freedom (DOF): sternoclavicular joint (2 DOF), glenohumeral joint (3 DOF),
scapulothoracic joint (4 DOF).
- Constraint point AC for the Acromioclavicular Joint.

2.2.1 Creation of supporting objects

The model was created from the dataset BodyParts3D, each surface of bone defining the body segments
was imported as STL file format. Anatomical landmarks were created that included the following
landmarks palpated on the bone surfaces: IJ, C7, PX, T8, on the thorax; SC, ACc, on the clavicle; ACs,
AA, TS, AI on the scapula; GH, EM, EL on the humerus (Figure 1). These landmarks are available in
NMSBuilder as sets of landmarks to create joint reference systems according to the ISB recommendations
[13].

(A) (B)
Figure 2: (A) Bony landmarks and local coordinate systems of the thorax, clavicle, scapula, and humerus
[17], (B) Thorax, clavicle, scapula, and humerus joint reference systems.
Thorax, clavicle, scapula, and humerus joint reference systems were created as anatomical reference
systems from the defined sets of landmark clouds. Two reference systems per joint are automatically
created, whose location and orientation are calculated with respect to the reference systems of the parent
body and the child body respectively, according to the ISB recommendations [17].

2.2.2 Creation of OpenSim objects and model

The OpenSim model objects were then created as follows:
- Bodies: all bodies were assigned the same density values, i.e. 1.42 g/cm3 and 1.03 g/cm3 for bone
and soft tissue surfaces [1], respectively. The inertial properties are automatically calculated by the
software and stored in the OpenSim model.
- Free joint with Ground: the thorax body was connected to the ground via a free joint.
- Joints: The sternoclavicular joint was defined as a ball joint type, which enables protraction-
retraction and elevation-depression of the clavicle since axial rotation cannot be accurately measured
and the conoid ligament limits axial rotation. The glenohumeral joint was defined as a custom gimbal
joint using the ISB standard coordinates for the angle of the elevation plane, elevation, and internal
rotation. Finally, the acromioclavicular joint was modeled as a ball joint by the constraint point AC.
- Marker Sets: all skeletal landmark clouds were attributed as marker sets to the corresponding bodies.
Finally, the created OpenSim model was exported in the OpenSim file format (.osim).

2.3. Scapulothoracic Joint Definition

The scapulothoracic joint was characterized by the translation and rotation of the scapula on the surface of
the thorax modeled as an ellipsoid [14]. The origin of the joint frame on the thorax was defined by the
center of the ellipsoid and the origin of the joint frame on the scapula was specified as the centroid of the
Angulus Acromialis (AA), Trigonum Spinae (TS), and Angulus Inferior (AI) markers, which are the
anatomical landmarks recommended by the ISB and the orientation was also computed according to the
ISB recommendations; however, the scapulothoracic joint frame is rotated -90° about the scapular Y-axis
to define upward rotation as a positive rotation about the scapulothoracic joint frame’s Z-axis.
 Figure 5: Scapulothoracic joint definition (OpenSim Model).

2.4. Scaling and Inverse Kinematics Simulation

2.4.1 Scaling model

The aim of scaling a generic musculoskeletal model is to modify the anthropometry of the generic model
so that it matches the anthropometry of a particular subject. We scaled upper limb segment dimensions to
obtain a correspondence between model markers and experimental marker locations identifying the same
bony landmarks. Scaling functionality was provided by using the scale tool in OpenSim. Scaling is one of
the most important steps in solving inverse kinematics and inverse dynamics problems because these
solutions are sensitive to the accuracy of the scaling step.

2.4.2 Inverse Kinematics Simulation

The purpose of inverse kinematics is to find the coordinates (joint angles) of the scaled model that best
reproduces the experimental kinematics. The inverse kinematics tool in OpenSim goes through each
recorded motion and computes the set of joint angles that put the model in a configuration that "best
matches" the experimental kinematics. We evaluated the kinematic accuracy of the model’s joints by
comparing thorax, scapula, and humerus model marker locations to measured marker data from bone-pin
motion-capture experiments [10]. Also, we evaluated the Euler angles for the scapulothoracic joints from
the experimental data and the model constructed according to the ISB recommendation. Ludewig et al.
measured marker locations with magnetic sensors fixed to the thorax on the anterior aspect of the
sternum, rigidly attached into the scapula at the scapular spine at the acromial base, into the lateral third
of the clavicle, and distal to the deltoid insertion on the lateral aspect of the humerus, for three tasks: arm
flexion, arm abduction and internal/ external rotation of the upper arm when the humerus is abducted at
the glenohumeral joint by 90° with respect to the thorax.

3. RESULTS AND DISCUSSION

After scaling the generic shoulder model, the model markers were in excellent agreement with
experimental markers in the static pose for each task, the root-means squared (RMS) marker errors were
less than 1 cm and all the maximum marker errors were less than 2 cm as shown in Table 1.
Table 1. Scaling Rusltats.
Total scuared RMS marker Maximum marker Marker with max
Task
error in cm error in cm error in cm error
Abduction 0.2522 0.5614 0.8106 PX
Flexion 0.4110 0.7168 0.8976 GH
Rotation 0.7113 0.9429 1.4814 EL
Table 2. Root-Mean Squared (RMS) in mm from three shoulder activities.

Task IJ T8 C7 AA AI TS GH EL EM Mean Worst

Abduction 6.12 2.85 6.22 4.91 7.19 5.03 7.59 8.15 7.35 6.16 EL
Flexion 2.13 2.81 2.70 3.83 5.83 2.89 7.32 1.89 1.65 3.45 GH
Rotation 1.58 1.49 3.64 4.41 2.43 3.83 7.21 2.68 2.53 3.31 GH
 (A) (B)
Figure 3: (A) Joint angles for all activities, (B) Euler angles for Scapula coordinates from the
experimental dataset and the current model.
The inverse kinematics of the joint were in excellent agreement with bone-pin measurements, which
means the model produced the kinematics from surface markers to within the accuracy of bone-pin
measurements (Table 2), which serves to validate the kinematics of the model. The joint angles against
time during the upper limb movement are shown in (Figure7-(A), left panel). The variability in the angle
of upper limb joints during activity performance between our model and Seth et al.’s model can be
observed in Figure 7-(A). whereas the collected motion dataset was the same, the graphs of upper limb
joints against time are variable, but the shape is similar. The most parameters that provoke this variability
are the bone morphology and the anatomical landmarks [12], which the last one must be performed
according to strict anatomical definitions to achieve higher reproducibility [9], This explains why
different results are found, partly because the anatomical coordinate systems used to represent the
kinematic results are dissimilar.

The kinematics resulting from the scapulothoracic joint in ISB standard rotation sequence (Figure7-(B))
was similar to the kinematics reported by Seth et al. [13], which means that the model reduced the
variability in Euler angles compared to angles computed directly from markers.

4. CONCLUSIONS

The upper limb movements kinematics by means of experimental data were analyzed by the OpenSim
model, which was created with the scapulothoracic joint using NMSBuilder software and the OpenSim
plugin for scapulothoracic joint definition. The results from this study showed that the shoulder model
produced the kinematics from surface markers to within the accuracy of bone-pin measurements, which
serves to validate the kinematics of the model. The similarity of Euler angles using the scapulothoracic
joint model to compute ISB-recommended angles of scapula orientation indicates that the model
constructed was in excellent agreement with Seth et al.’s model. however, the joint angles were
dissimilar, because the anatomical coordinate systems used to represent the kinematic results were
different. In conclusion, our model can be used for inverse and dynamics analyses and to compute joint
reaction loads.

REFERENCES

01- Charlton IW, Johnson GR A model for the prediction of the forces at the glenohumeral joint. Proc
Inst Mech Eng H (2006).
02- Damsgaard M, Rasmussen J, Christensen ST, et al: Analysis of musculoskeletal systems in the
AnyBody Modeling System. Simul Model Pract Theory (2006).
03- De Baets L, Van Deun S, Desloovere K, Jaspers E: Dynamic scapular movement analysis: Is it
feasible and reliable in stroke patients during arm elevation? PLoS One (2013).
04- Della Croce U, Leardini A, Chiari L, Cappozzo A. Human movement analysis using stereo
photogrammetry. Part 4: assessment of anatomical landmark misplacement and its effect on joint
kinematics. Gait Posture. (2005).
05- Garner BA and Pandy MG. A kinematic model of the upper limb based on the visible human project
(VHP) image dataset. Computer Methods and Programs in Biomedicine (1999).
06- Giordano Valente, Gianluigi Crimi, Nicola Vanella, Enrico Schileo, Fulvia Taddei, NMSBuilder:
freeware to create subject-specific musculoskeletal models for Open- Sim, Computer Methods and
Programs in Biomedicine (2017).
07- H. Gill, L. Gustafsson, L. Hawcroft, and K. McKenna, “Shoulder joint range of motion in healthy
adults aged 20 to 49 years,” Br. J. Occup. Ther (2006).
08- Katherine R. S. Holzbaur, Wendy M. Murray, Scott. Delp: A Model of the Upper Extremity for
Simulating Musculoskeletal Surgery and Analyzing Neuromuscular Control. Annals of Biomedical
Engineering (2005).
09- Katherine R. S, Xiao Hu, Craig M. Goehler, Meghan E. Vidtd, Melissa Daly, Anca Velisar & Wendy
M. Murray: Benchmarking of dynamic simulation predictions in two software platforms using an
upper limb musculoskeletal model. Computer Methods and Programs in Biomedicine (2015).
10- Ludewig PM, Phadke V, Braman JP, Hassett DR, Cieminski CJ, LaPrade RF: Motion of the shoulder
complex during multiplanar humeral elevation. J Bone Joint Surg Am (2009)
11- Nikooyan AA, Veeger HE, Chadwick EK, et al. Development of a comprehensive musculoskeletal
model of the shoulder and elbow. Med Biol Eng Comput (2011).
12- Serge Van Sint Jan: Introducing Anatomical and Physiological Accuracy in Computerized
Anthropometry for Increasing the Clinical Usefulness of Modeling Systems. Critical Reviews in
Physical and Rehabilitation Medicine (2005).
13- Seth A, Ricardo Matias, António P. Veloso, Scott L. Delp: A Biomechanical Model of the
Scapulothoracic Joint to Accurately Capture Scapular Kinematics during Shoulder Movements. PLoS
One (2016)
14- Seth A, Sherman M, Eastman P, Delp S Minimal formulation of joint motion for biomechanisms.
Nonlinear Dyn (2010)
15- Seth A, Jennifer L. Hicks, Thomas K. Uchida, Ayman Habib: OpenSim: Simulating musculoskeletal
dynamics and neuromuscular control to study human and animal movement. PLoS One (2019).
16- Srikumaran U, Wells J, Freehill M Scapular Winging: A Great Masquerader of Shoulder Disorders. J
Bone Jt (2014)
17- Wu G, van der Helm FCT, Veeger HEJ, Makhsous M, Van Roy P, Anglin C, et al: ISB
recommendation on definitions of joint coordinate systems of various joints for the reporting of
human joint motion—Part II: shoulder, elbow, wrist, and hand. J Biomech (2005).