Prediction Of Normal Glenoid Morphology With Statistical Shape Modeling

Berger, Steve; Terrier, Alexandre; Becce, Fabio; Farron, Alain; Büchler, Philippe (2016). Prediction Of Normal Glenoid Morphology With Statistical Shape Modeling (Unpublished). In: European Federation of National Associations of Orhtopaedics and Traumatology. Geneva.

Introduction
Patients suffering from glenohumeral osteoarthritis may present an important wear of the glenoid
articular surface leading to a new glenoid shape (Neoglenoid). The amount of wear to correct can thus
be difficult to evaluate when a total shoulder arthroplasty is planned. Since the positioning of the
glenoid component is known to be critical for the long term success of the surgery, it would be helpful
to have an estimate of the glenoid shape before it was deformed by osteoarthritis (Paleoglenoid).
Objectives
The objective of this project was to evaluate the potential of statistical shape modeling to predict the
shape of the articular glenoid surface from the remaining parts of the scapula, which have not been
affected by the glenoid degenerative wear.
Methods
To build a statistical shape model (SSM) of the scapula, CT scans of 64 scapulae without any sign of
pathology were segmented. Segmented images were used to produce smooth three-dimensional surfaces
of triangular meshes. The 64 meshes were then registered and combined into a SSM using the Statismo
library. To evaluate the accuracy of the SSM prediction, we repeated the following six steps for each of
the 64 scapulae. Step 1: one of the 64 scapulae is selected. Step 2: the SSM is built without this scapula
(leave-one-out). Step 3: the glenoid region of the selected scapula is removed. Step 4: the SSM is fitted
based on the remaining part of the scapula. Step 5: the SSM predicts the removed glenoid region. Step
6: the local accuracy of the SSM is evaluated by the distance between the predicted and existing glenoid
surfaces. We compared a SSM with the full scapulae to a SSM with the lateral part only (from the
spino-glenoid notch to the acromion and coracoid process). The orientation of the original and
reconstructed glenoid were also assessed.
Results
When the SSM included the entire scapula, the lowest prediction accuracy of the reconstructed glenoid
surface reached 4 mm. When the SSM was based on the lateral scapula, the average accuracy reached
0.6 ± 0.1 mm. The accuracy of the glenoid orientation prediction was 2.4 ± 1.6 degree.
Conclusions
The proposed method allowed to predict the shape of the glenoid surface, which was here artificially
removed. This prediction could be obtained from the lateral part of the scapula, glenoid excluded,
which is not affected by wear caused by osteoarthritis. The accuracy of the reconstruction was below 1
mm, and thus acceptable for the preoperative planning of total shoulder arthroplasty. This method
should help to improve the positioning of glenoid implants for severe glenoid wear, and eventually
reduce the failure risk after total shoulder arthroplasty.

Item Type:

Conference or Workshop Item (Speech)

Division/Institute:

10 Strategic Research Centers > ARTORG Center for Biomedical Engineering Research > ARTORG Center - Computational Bioengineering
04 Faculty of Medicine > Pre-clinic Human Medicine > Institute for Surgical Technology & Biomechanics ISTB [discontinued]

UniBE Contributor:

Berger, Steve, Büchler, Philippe

Subjects:

500 Science > 570 Life sciences; biology
600 Technology > 610 Medicine & health

Language:

English

Submitter:

Steve Berger

Date Deposited:

23 Mar 2017 15:07

Last Modified:

01 Jul 2024 10:33

URI:

https://boris.unibe.ch/id/eprint/95750

Actions (login required)

Edit item Edit item
Provide Feedback