Evaluation of bone impingement prediction in pre-
operative planning for shoulder arthroplasty
P R Krekel
1,2
*
, P W de Bruin
3
, E R Valstar
1,4
, F H Post
2
, P M Rozing
1
, and C P Botha
2,3
1
Department of Orthopaedics, Leiden University Medical Center, Leiden, The Netherlands
2
Data Visualisation Group, Delft University of Technology, Delft, The Netherlands
3
Department of Radiology, Leiden University Medical Center, Leiden, The Netherlands
4
Department of Biomechanical Engineering, Delft University of Technology, Delft, The Netherlands
The manuscript was received on 14 October 2008 and was accepted after revision for publication on 19 December 2008.
DOI: 10.1243/09544119JEIM531
Abstract: In shoulder arthroplasty, malpositioning of pros-theses often leads to reduced post-
operative range of motion (ROM) and complications such as impingement, loosening, and
dislocation. Furthermore, the risk of impingement complications increases when reverse total
prostheses are used. For this purpose a pre-operative planning system was developed that
enables surgeons to perform a virtual shoulder replacement procedure. Our pre-operative
planning system simulates patient-specific bone-determined ROM meant to reduce the risk of
impingement complications and to improve the ROM of patients undergoing shoulder
replacement surgery. This paper describes a validation experiment with the purpose of ratifying
the clinical applicability and usefulness of the ROM simulation module for shoulder
replacement surgery.
The experiment was performed on cadaveric shoulders. A data connection was set up
between the software environment and an existing intra-operative guidance system to track the
relative positions of the bones. This allowed the patient-specific surface models to be visualized
within the software for the position and alignment of the tracked bones. For both shoulders,
ROM measurements were recorded and tagged with relevant information such as the type of
prosthesis and the type of movement that was performed. The observed ROM and occurrences
of impingement were compared with the simulated equivalents. The median deviation
between observed impingement angles and simulated impingement angles was 20.30u with an
interquartile range of 5.20u (from 23.40u to 1.80u). It was concluded that the ROM simulator is
sufficiently accurate to fulfil its role as a supportive instrument for orthopaedic surgeons during
shoulder replacement surgery.
Keywords: medical visualization, pre-operative planning, shoulder, arthroplasty, range of
motion, motion tracking
1 INTRODUCTION
Shoulder arthroplasty (Fig. 1) aims to provide pain
relief and to restore joint mobility. However, success
rates in shoulder replacements are considerably
lower than in knee and hip replacements [1]. This
can be partially ascribed to the complexity of the
shoulder joint. The surgical exposure provides little
room for manoeuvring the instruments and causes the
field of view for the surgeon to be limited. This
frequently results in malalignment of prostheses [2].
Malalignment of shoulder prostheses leads to a
diverse range of complications, e.g. increased wear,
reduced stability of the glenohumeral joint, limited
range of motion (ROM), and impingement [3, 4].
Long-term consequences of impingement are abra-
sion of bone, loosening of the prosthesis, and
increased risk for the necessity of revision surgery [5].
Impingement complications are frequently seen
with reverse shoulder prostheses, a type of prosthesis
*Corresponding author: Department of Orthopaedics, Leiden
University Medical Center, Albinusdreef 2, Leiden, 2333ZA, The
Netherlands. email: P.R.Krekel@lumc.nl
SPECIAL ISSUE PAPER 1
JEIM531 F IMechE 2009 Proc. IMechE Vol. 223 Part H: J. Engineering in Medicine

usually indicated for patients who suffer from a
severely impaired rotator cuff [6–8]. Scapular
notching refers to impingement of the humeral
cup with the glenoid neck and is the type of
impingement that is most common for reverse
shoulder prostheses. Sirveaux et al. [9] carried out a
multi-centre study consisting of 80 shoulders,
which showed scapular notching in 51 cases. A
similar study by Juvenspan et al. [10] showed
scapular notching in 39 of 55 cases. Conversely,
several studies show that careful planning greatly
reduces the risk on scapular notching [11–13].
Anticipating the frequent occurrence of impinge-
ment and the clear need to prevent it, a pre-
operative planning system for shoulder replace-
ments that interactively simulates bone-deter-
mined ROM was developed. The system was
described in detail in a previous publication [14].
A short summary of the system is included in
section 2.1.
This paper describes an experiment for the
validation of the ROM simulations. The purpose
of the experiment was to determine the accuracy of
the ROM simulator. In addition, the ability of
the system to predict impingement complications
such as scapular notching was evaluated. The
prediction of bone impingement enables the sur-
geon to set up a plan that explicitly avoids this type
of complication.
2 METHODS
2.1 Description of the pre-operative planning
system
The pre-operative planning system loads computed
tomography (CT) data and extracts surface models of
the scapula and humerus. Subsequently, the surgeon
can place prosthesis models on the bone models.
Currently, the system only supports total glenohum-
eral prostheses, i.e. both a cup and a ball component
have to be placed. The system does support total
reverse prostheses.
The system automatically calculates the position
of well-known landmarks on the bone models.
Planes are moved through the object to find the
most extreme points on the models. The centre of
rotation of the glenoid is determined by applying a
Hough transform to the surface models. The land-
marks can also be selected manually. Using the
landmark positions, a predefined pre-operative plan
is transformed to the patient-specific case. Subse-
quently, prostheses can be dragged to different
positons, thereby altering the plan.
In order to calculate the ROM, a biomechanical
model of the glenohumeral joint was implemented. A
generally accepted hypothesis is that the glenohumeral
joint can be approximated by a ball joint [15, 16].
No displacement is taken into account. The
scapula coordinate system as described in the
Fig. 1 Illustration of a shoulder joint after total shoulder replacement
2 P R Krekel, P W de Bruin, E R Valstar, F H Post, P M Rozing, and C P Botha
Proc. IMechE Vol. 223 Part H: J. Engineering in Medicine JEIM531 F IMechE 2009